Prof. Shweta Agrawal, CSE | "Real-life cryptography is cooler than Imitation Game-the movie"| Ep.13

Shweta Agrawal 0:00 Calling our field computer science is like calling surgery knife science. Cryptography, in a sentence, is the art of keeping secrets. The desire to keep secrets is as old as humanity itself. Cryptography is really this very fine dance between sort of structure and randomness. I remember reading this algorithm as an undergrad and feeling that this is a life-changing moment. [upbeat music]

Unknown Host 0:38 Hi, my name is Amrit. We've heard that IIT Madras is the best place to build. [upbeat music] So we've come down to the Sudha and Shankar Innovation Hub. We want to meet some people. These are builders. We want to talk to them about their work, and also ask them, what makes IIT Madras the best place to build? [upbeat music] Hi, welcome to the Best Place to Build Podcast. Today, we are sitting with Professor Shweta Agrawal. She's a professor of computer science at IIT Madras. Her area of interest is cryptography, so I've brought my ruled notebook to take notes. Uh, she holds the Venky Hariharan and Anand Rajaram Chair, and has won numerous awards, including the Swarnajayanti in 2020. Welcome, ma'am.

Shweta Agrawal 1:32 Thank you so much. It's good to be here.

Unknown Host 1:34 Ma'am, let's start with,

Unknown Host 1:35 uh, what is cryptography? I think, uh, maybe it's obvious in your world, but it's completely alien to me.

Shweta Agrawal 1:42 So cryptography, in a sentence, is the art of keeping secrets. Uh, essentially, there are different goals that you might have, uh, such as, you know, wanting to communicate with someone in private, or when you're talking to someone, you, you may want to convince them that, you know, you are the person that they think, uh, they are talking to. So there are different goals that, uh, you might want to achieve in a secret manner, right? And cryptography is the science of enabling that.

Unknown Host 2:10 So what I know of this, from whatever you said, is, uh... So I've watched Imitation Game, which is a movie about, uh-

Shweta Agrawal 2:16 Yeah

Unknown Host 2:17 ... about Alan Turing-

Shweta Agrawal 2:18 Right

Unknown Host 2:18 ... his team breaking the Enigma code. So c- can you tell us where we started from in cryptography and where we have

Unknown Host 2:24 come to?

Shweta Agrawal 2:24 Yeah, so cryptography, uh, is really an ancient field, though it may not have been, you know, uh, given, given this name until much more recently. But it's, uh... You know, the, the desire to keep secrets is as, uh, old as humanity itself, right? So y- you have, like, these very, very old codes, like the Caesar cipher and so on, where essentially if I want to say something in private to you, I replace each letter by, you know, the letter five steps ahead of it or something like that. So, uh, the field in, in its original form, I think has always existed, as long as humanity has existed. Uh, the movie you mentioned, and I mean, uh, the, the war and the efforts of building, uh, unbreakable codes, of course, took on, uh, like, a profound importance, uh, during that time. And, uh, the field as we know it is actually also, in a way, inspired from what happened, uh, during the war, which is that, uh, you know, a, a bunch of really smart people put together a really, uh, hard code to break, what they thought was a hard code to break, but they missed something, right? And that's what led to the, the break of the code. So modern cryptography seeks to kind of address this problem by saying that, "You know, we will prove to you mathematically that the only way that you can break our code is by solving some, you know, very hard mathematical problem." So in some sense, you know, all attack vectors can be reduced down to, like, this one question of: Can you solve so and so problem? And then if you choose this underlying problem as something that, you know, maybe mathematicians have tried for, like, decades, uh, and you have a lot of confidence that this problem is going to be very hard for any attacker, then you can have much more confidence that your code is secure.

Unknown Host 4:14 Okay.

Shweta Agrawal 4:14 So that is the field of modern cryptography.

Unknown Host 4:16 I understood. So let me just say that, for example,

Unknown Host 4:20 if I use a Caesar cipher, then the code to break is that I have to figure out that this is five letters ahead, and I just have to go back five letters. But we have come to a place where figuring out how to break this code has become much harder.

Shweta Agrawal 4:33 Yeah.

Unknown Host 4:33 And I'll, I'll ask you in a bit about hard... What do you mean by hard? Because I think what I mean by hard is in sense of the word hard, but maybe you are using a more mathematical, uh, definition. But before that, in the movie, and, and because we started from there, and because you said that a lot of it started from the war effort, in the movie, um, the Germans deve- developed this code, uh, and the, uh, Allied powers are trying to break it. It- is it that... Uh, I mean, it feels like a national security kind of issue, but why do we, as normal people or normal businesses, why should we

Unknown Host 5:09 care about cryptography?

Shweta Agrawal 5:10 So first of all, I mean, if you lose wars, uh, as citizens- [chuckles]

Unknown Host 5:13 Everybody has to. [chuckles]

Shweta Agrawal 5:14 ... everybody has to care. But, uh, that- that's a good question you raised, because obviously, uh, it has a lot of real-world implications, even outside of defense considerations. So for instance, you know, all of us are doing online shopping now. Now, when you do online shopping, you use your credit card information, right? And this credit card, uh, number and the CVV code, et cetera, and the OTP, this is all traveling over the network. So if there's an eavesdropper who's listening in to this information, right, uh, she can intercept it and use it to make other purchases of, of her own. Well, not the OTP, but for instance, if, if it's just the card number and CVV.... so, uh, you need to encode this information for it to be secure, even to travel over the network, right? So this is a very basic example of where cryptography is used in day-to-day life, but there are lots of others. Like, let's say that, you know, two companies want to try, uh, want to figure out whether they, they should merge or not. Will it be profitable? Now, before the merger, you don't want to reveal your assets to the other company, but, uh, computing whether the merger is going to be profitable requires information about the assets. So one way that cryptography can enable this is that, you know, uh, you can run a, a so-called multiparty computation protocol-

Unknown Host 6:34 Okay

Shweta Agrawal 6:35 ... where, uh, the two of us have private inputs, right? And we're interested in computing some common function, and there's a mathematical guarantee that, uh, each party will only learn the output of the function, but nothing else about the other person's input. So it has really very wide applications.

Unknown Host 6:53 Fair enough. I guess, um, if I'm... I mean, just reducing it back to the personal, if I want to share with a bank that I'm worthy of certain financial-

Shweta Agrawal 7:02 Yeah

Unknown Host 7:02 ... investment, but don't want to divulge my assets?

Shweta Agrawal 7:04 That's right.

Unknown Host 7:05 Okay. Great. So

Unknown Host 7:07 you used the word eavesdropper, and then I think earlier you had used the word attacker.

Shweta Agrawal 7:11 Yeah.

Unknown Host 7:11 Are these terms that are used often in this field?

Shweta Agrawal 7:14 These are terms that, uh... yeah, they, they, uh, they get used in almost every sentence. But, uh, when, when we're thinking of these terms from the cryptographic perspective, really, these are not people, but algorithms. So, you know, we're, we're really thinking of an algorithm which takes some information and, you know, can it do something useful with that information? So-

Unknown Host 7:36 Okay. Okay.

Shweta Agrawal 7:36 Yeah.

Unknown Host 7:36 So are these the same, by the way, this eavesdropper and attacker?

Shweta Agrawal 7:39 No, they are not the same. So an eavesdropper is a particular type of attacker who, who passively listens on a channel, but an attacker could be able to do more. So, for example, an attacker, uh, could be an algorithm that not only listens, but also somehow changes what is being sent on the channel, right? So all kinds of attacks are possible.

Unknown Host 7:59 Fabulous. So let's go back

Unknown Host 8:01 to the word hard. Uh, you used the word, um, hard problem.

Shweta Agrawal 8:05 Yeah.

Unknown Host 8:05 I assume that there's some definition to that.

Shweta Agrawal 8:07 Yeah. So, uh, you know, at a high level, by hard, what we mean in computer science is that no efficient probabilistic algorithm should be able to solve this problem. And by efficient, you know, uh, polynomial time is, is the technical term that we use. But, uh, for all practical purposes, this just means that, you know, your, your fastest supercomputer should not be able to solve this problem.

Unknown Host 8:32 Okay.

Shweta Agrawal 8:32 All of them put together also should not be able to solve this problem in any reasonable time. And the reasonable time here, depending on your application, it could be 10 years, like, if it is just, you know, some very local application, or it could be, you know, the age of humanity. So you, you can kind of vary the kind of security that you ask from the scheme, uh, and, and set the hardness parameter accordingly.

Unknown Host 8:57 It's very interesting. Again, going back to the movie, because that's really [chuckles] my only reference point to cryptography, uh, what happens is that-

Shweta Agrawal 9:04 I just want to say that real-life cryptography is cooler than this movie.

Unknown Host 9:08 [chuckles] Oh, fair enough. But to my point, there was a day limit, that the code changes every day.

Shweta Agrawal 9:13 Yeah.

Unknown Host 9:13 So they had to s- they had to sort of race to solve it-

Shweta Agrawal 9:15 Yeah

Unknown Host 9:15 ... within a day. I want to just dig deeper a little bit onto this, because maybe I don't understand it as well. I have a very superficial understanding of the word algorithm

Unknown Host 9:24 itself. Uh, so what does an algorithm mean in a theoretical sense?

Shweta Agrawal 9:30 An algorithm is, uh, you know, just a sequence of steps that will solve some problem for you. So there, there is some problem, right? There is some input, and there is a specification of what output you want.

Unknown Host 9:41 Mm.

Shweta Agrawal 9:42 And, uh, the algorithm is the sequence of steps that tells you how to go from the input to the output.

Unknown Host 9:47 Okay, and what would be an example of an algorithm that's used in encryption or decryption?

Shweta Agrawal 9:53 You know, multiply a matrix with a vector.

Unknown Host 9:56 Okay.

Shweta Agrawal 9:56 Uh, so the algorithm would be sort of, you know, uh, compute the inner product of the, the first row of the matrix with the vector, and then go to the next one. Maybe this is, uh, too superficial, uh, an example.

Unknown Host 10:10 I've heard

Unknown Host 10:11 this word RSA, AES. What are these terms?

Shweta Agrawal 10:14 Yeah, so I was just coming to that. So those are kind of more interesting examples of algorithms that are used in cryptography. The matrix vector is, is like a toy example. Uh, RSA is a very, very famous cryptosystem. Uh, it, it won its inventors the Turing Award, and it's actually the reason- the place where I fell in love with cryptography. Like, I, I remember reading this algorithm as an undergrad and, you know, feeling that this is a life-changing moment. So it's this very clever algorithm which, uh, shows you how to take, uh, some message and, you know, convert it into a ciphertext, what we call a ciphertext. So the input is the message, the output is the ciphertext. There's also some random coins that the algorithm flips along the way, and it does a sequence of operations that will convert the message into something that is unbreakable. So now the ciphertext, if I send it on the channel, you can, you know, uh, read it, you can try everything you want, but unless you can solve some very hard problem which is considered infeasible, you cannot reverse map it back to the message that you started with.

Unknown Host 11:21 Okay. Okay, but, but somebody will be able to reverse, because it has to be decrypted, right?

Shweta Agrawal 11:27 Very good. So that's very important. So, uh, the point is that, you know, uh, RSA is a, what, what we call a public key encryption scheme. So here there, there is a public key and a secret key, and, uh, in order to encrypt, you only use the public key, so anybody can encrypt. So this encryption algorithm, anybody can run it, but the decryption algorithm needs a secret, uh, input which, which is the secret key. So now this kind of reduces your task to just keeping the secret key secure. So if the secret key is secure, then, uh, you know, your ciphertext is, uh-... uh, unbreakable, but of course, if you have the secret key, then you can get back the message.

Unknown Host 12:07 As you're saying this, I remember, uh, uh, some memes, internet memes around people sharing their private key or public key on GitHub or something like that. So that, that's- so I would generate a private key and a public key. I would share the public key with the, the people-

Shweta Agrawal 12:22 Users

Unknown Host 12:22 ... I'm sharing with.

Shweta Agrawal 12:23 Yeah.

Unknown Host 12:23 And give my private key to myself. That's the-

Shweta Agrawal 12:25 Yeah

Unknown Host 12:26 ... idea. And so then I don't need to worry about what the RSA algorithm actually does, I just have to keep the private key.

Shweta Agrawal 12:30 That's right. Yeah.

Unknown Host 12:32 Mm. Okay. Uh, so, um, uh, so far we have spoken about what is cryptography, P versus NP. Polynomial time is what you said, and, uh, what is a hard problem, and some algorithms we use.

Shweta Agrawal 12:43 So we didn't

Shweta Agrawal 12:44 actually talk about P versus NP, but this is, this is, uh, at the root of the, the hardness, uh, question. So P versus NP is the most famous problem of computer science, and, you know, uh, it, it's still open. So it, it's something that shows us, uh, how little we as humanity know, uh, about science, right? So it's a very, very basic question. P is, is the class of pro- problems that can be solved efficiently. NP is the class of problems whose solution can be verified efficiently. Now, obviously, if you can solve a problem efficiently, you can also check the answer efficiently just by solving it, right? But, uh, there, there can be problems which you don't know how to solve, but if you're given the answer, you can check if it's the right answer. And the question is: Are these two classes the same? So cryptography, at the very least, uh, requires that these two classes be different, because you, you should be able to base the hardness of, of your, uh, uh, s- scheme, like the security of your scheme, on some problem that cannot be solved efficiently.

Unknown Host 13:50 So if P is a set of problems and NP is another set of problems, there should be a problem in NP which is not in-

Shweta Agrawal 13:54 Exactly. Exactly.

Unknown Host 13:55 And do, do problems move from here to there over time?

Shweta Agrawal 13:59 Um, yeah, that's a good question. So yes, I mean, um, uh, there, there are problems for which, you know, we find polynomial time algorithms which were not known before. So certainly, yeah.

Unknown Host 14:11 Oh, so the P is polynomial time?

Shweta Agrawal 14:13 Yeah, P-

Unknown Host 14:13 And why is it called polynomial times?

Shweta Agrawal 14:15 Because, uh, technically speaking, the- when I say efficient, what we mean is that, uh, the running time of the algorithm is polynomial in, in the size of the problem. So if the problem took n bits to describe, right, then the time to solve the problem should be n to the c for some constant c. That's polynomial time.

Unknown Host 14:36 As opposed to exponential, or e to the...

Shweta Agrawal 14:38 Exponential is, is, uh, yeah, beyond polynomial but, you know, there are also functions in the middle. But, yeah, you could think of exponential as being hard. Like, if, if you think that a problem will take exponential time to solve, this is a hard problem.

Unknown Host 14:54 Does the invention of quantum computers move more problems from NP to P? Is that-

Shweta Agrawal 15:00 So actually, P and NP, the way that we define them are with respect to classical computers. I mean, you can define these problem, uh, these classes, uh, correspondingly in the quantum setting. And, uh, I think, uh, the short answer to what you're really asking is that, yes, there are problems that were classically, uh, considered hard that became easy in the quantum setting. And, uh, RSA, which we talked about earlier, is, is the most famous such example. So, so breaking an RSA, uh, cryptosystem is something that we believe is hard for classical computers. I say believe because remember, uh, I said that whether P and NP are the same is itself something that is unknown. Like it, it's conjectured, but, you know, we don't have a proof. So you can only conjecture that some problem is hard. And RSA is, uh, believed to be hard based on, you know, decades of research, uh, for classical computers. But for quantum computers, it- it's actually easy to solve.

Unknown Host 16:05 Mm. So a physical realization of quantum computers would sort of change the field entirely?

Shweta Agrawal 16:12 I mean, the field has already changed entirely because for us, you know, the physical realization doesn't matter. Like, the algorithm is the physical realization. I mean, for, for-

Unknown Host 16:20 For theoretical computer scientists.

Shweta Agrawal 16:21 For the- yeah, for theoretical computer scientists, uh, the algorithm is somehow more real than the computer, right? Because the, the computer may or may not be there, but the algorithm is already there. So you, you already care about, uh, using algorithms that, uh, don't have efficient quantum alga- uh, quantum solvers.

Unknown Host 16:41 Okay. So the, the description of a quantum computer already is there, so you already know that-

Shweta Agrawal 16:46 Yeah

Unknown Host 16:46 ... uh, it can-

Shweta Agrawal 16:48 Exactly

Unknown Host 16:48 ... it can solve RSA in polynomial time.

Shweta Agrawal 16:50 Yeah.

Unknown Host 16:51 The physical realization of the quantum computer, I-

Shweta Agrawal 16:55 May come now, may come later, but the point is, you know, it's something that we have to worry about now.

Unknown Host 17:00 Right. So if I could buy a quantum computer off Amazon, [chuckles] then I could theoretically break... I mean, practically break RSA algorithm.

Shweta Agrawal 17:07 Yeah, I mean, this is extremely far away-

Unknown Host 17:09 Yeah

Shweta Agrawal 17:09 ... uh, if at all, but, uh, sure. Yes. The answer is yes.

Unknown Host 17:14 Okay, understood. Um, with this background, can you

Unknown Host 17:18 give us a little bit of idea on your own research interests and your-

Shweta Agrawal 17:22 Yeah

Unknown Host 17:22 ... work?

Shweta Agrawal 17:22 So I work in, uh, theoretical cryptography, and this, as, uh, I've been describing, is, is, you know, the science of designing algorithms that are secure to do interesting tasks, right? And, uh, the most basic task in cryptography is of encryption, and it's actually deep enough that, you know, you can sort of, uh, spend all your time just researching that. In, in a way, I have done that. Because encryption is not just about, uh, keeping data private. There are all kinds of other very interesting questions that you can ask about encryption. So, for example, if

Shweta Agrawal 17:57 I want to, uh, encrypt my data, but I still want to do some meaningful computation on the data. Like, so for example, machine learning is a big thing now, where, you know, you have data and you want to run some algorithm and learn some- something at the end of it.... but in order to run this algorithm, everybody needs to share their data, right? But you may not trust the, the person who's collecting the data and, um, running this algorithm. So the question here that can be asked is, can you encrypt the data but still run your machine learning algorithm on it? And this is something which if you think about it for, for a bit, it seems like it should not be possible, right?

Unknown Host 18:39 Can I just, uh, the, the example that I have learned from your earlier talks is one of medical researcher-

Shweta Agrawal 18:46 Yes, yes

Unknown Host 18:46 ... looking at pe- a mass number of people's-

Shweta Agrawal 18:48 That's right

Unknown Host 18:49 ... health data.

Shweta Agrawal 18:50 For example, or genomic data.

Unknown Host 18:52 Genomic data.

Shweta Agrawal 18:53 So I often use this example because I think it's very relatable. Like, everybody will agree that, you know, medical research is one of the most important, uh, fields of research, right? And, uh, now you have a, uh, all this, uh, genomic data that could potentially be used for medical research. Like, maybe you can study, you know, how genomic patterns correlate with the, uh, efficacy of certain medication.

Unknown Host 19:17 Sure. For example, I want to solve for diabetes-

Shweta Agrawal 19:20 Yeah

Unknown Host 19:20 ... and I want to, like, ask for the genomic data of-

Shweta Agrawal 19:23 That's exactly

Unknown Host 19:23 ... 1 million diabetic patients.

Shweta Agrawal 19:24 Yeah. Exactly. So now the point is that, uh, these 1 million diabetic patients are, are, uh, not going to be excited about sharing their genomic data with anybody because, you know, genomic data is something very, very private and sensitive.

Unknown Host 19:39 Yeah.

Shweta Agrawal 19:39 So the question will be that can I encrypt it and still have you do the research you want?

Unknown Host 19:45 Sure.

Shweta Agrawal 19:46 And now this is, uh, something which is really like walking this fine line, uh, of, you know, possible and impossible sort of structure and

Shweta Agrawal 19:55 randomness. Because, uh, if you think about it, when I say that a cipher is unbreakable, what does this really mean? It means that, you know, whatever I saw... Let's say I'm an attacker, whatever I saw, this should look completely-

Unknown Host 20:10 Mm

Shweta Agrawal 20:10 ... random. It should look, you know, just like noise in some sense, right? Junk. It should look like junk. But now, um, what I'm saying is, i- if I want to compute, like if I want to add two numbers, okay, and I've encrypted them, I can add two numbers when they're given to me in the clear,

Shweta Agrawal 20:30 but once I encrypted them, it's like I put them into two boxes that look exactly the same. They have to look exactly the same because if they didn't, I could distinguish which number is in which, right? Think of bits just zero and one.

Unknown Host 20:43 Yeah.

Shweta Agrawal 20:43 So if both of them are put inside an, uh, encryption scheme, right, there, there's a ciphertext. This looks like a black box. Okay. Now, the, the shape and sort of... Think, think of a box. Like, the, the shape, size, and appearance of this box should not tell you anything about whether it's zero inside or one inside.

Unknown Host 21:03 Yeah.

Shweta Agrawal 21:04 So in some sense, you know, encryption is, is making everything look the same.

Unknown Host 21:09 Sure. Mm.

Shweta Agrawal 21:10 But now when I want to compute, let's say I want to add or multiply these numbers, right? I cannot, uh, do that if, if it's the same object.

Unknown Host 21:20 Sure, sure.

Shweta Agrawal 21:20 Right?

Unknown Host 21:21 Yeah.

Shweta Agrawal 21:21 So in some sense, you know, cryptography is really this very fine dance between sort of structure and randomness, where the randomness is, you know, that everything looks random. So there needs to be a computational hardness, which says I cannot recover zero and one given this box.

Unknown Host 21:41 Yeah.

Shweta Agrawal 21:42 But there needs to be somehow some hidden structure in this, uh, random-looking object that allows me nevertheless to add, multiply, and eventually decrypt.

Unknown Host 21:52 So this image I'm getting in my mind is, uh, that, um... I don't know if this is accurate, but I will share it anyway. Suppose, uh, you know, you've seen these large container ships which have, like, 1,000 containers, uh, and so they all look the same. Maybe the color is slightly different. One is orange, one is blue or something. So there's some data inside the container. I need to run a computation on what is inside, uh, but I should not be able to see it, and, uh, I can only see containers.

Shweta Agrawal 22:19 Yeah.

Unknown Host 22:20 But my algorithm can figure out whatever computation it needs without actually having access to the-

Shweta Agrawal 22:24 Exactly. Exactly. So it's like saying... So here I'm not even allowing different colors, right? They- all these containers look the same. Let's say, you know, they're, they're just all boxes of the same shape and size and the same color. Now what I'm saying is I take, you know, 5 million black boxes, okay? Each one has the genomic data of 5 million people. I do some magic, right, and I, I give you a, a box which contains the answer of whatever you wanted to compute.

Unknown Host 22:58 Sure.

Shweta Agrawal 22:59 And I manage to also give you a secret key, which opens this box in a way that the secret key can only open the box if the computation was done correctly. Like, if you did some illegal computation, which is to say, you know, you took 5 million boxes, but you said, "You know, the computation I'll actually do is just extract-

Unknown Host 23:21 Sure, gene sequence or a part

Shweta Agrawal 23:23 ... gene sequence of, of so and so person."

Unknown Host 23:24 Yeah.

Shweta Agrawal 23:24 Right? If this is the computation which you, you, you actually wanted to do, um, the key will not open that box.

Unknown Host 23:33 Okay.

Shweta Agrawal 23:33 So only if you did the correct computation, which is, you know, whatever maybe everybody agreed. So everybody agreed that, okay, uh, you know, since I'm diabetic, I, I contribute my data to studying, uh, you know, some- something that will help diabetes, right? Everybody signs off on that, and now I, I can give a key which only opens the computation that everybody signed off.

Unknown Host 23:56 Understood. So the encryption is specific to that, uh, request, that, that-

Shweta Agrawal 24:00 No, the encryption is not specific. The encryption is actually... It can be done. So that, that, that's a very good, uh, point because if the encryption were specific to that request, it would be a much easier problem. The encryption is just encryption. It, it, it's not married to any particular functionality, but I could, uh, you know, use, uh, different authorized functionalities-... to compute on the same ciphertext. Like, because a person is so many things, right? Like, so, uh, you know, maybe a person is diabetic, but you know, maybe she's also, I don't know, like, a ballet dancer or a professor or whatever. So you could be having a different study, let us say, that, you know, requires you to collect, uh, the data of all dancers together, and maybe she's also okay with that. So for each new study, she does not need to encrypt her data differently.

Unknown Host 24:54 Okay. I, I think- I feel like I understood [chuckles] that. Uh, is this what is called functional encryption?

Shweta Agrawal 24:58 This is called functional encryption, yes.

Unknown Host 25:01 Okay.

Shweta Agrawal 25:01 And this, this has been one of the core things that I have worked on, like designing algorithms to enable this.

Unknown Host 25:07 Okay. And-

Shweta Agrawal 25:08 And in particular, since we talked about, uh, quantum, right? Uh, in particular,

Shweta Agrawal 25:13 the solutions I care about are, uh, those that, uh, we think will be secure even in the quantum regime. So even if there are quantum algorithms that are trying to break open the ciphertext, even then these constructions should remain secure.

Unknown Host 25:30 I've also... I, I, I went through the, the page that, uh, the department maintains, um, on, on your work, and I noticed something called, uh, attribute-based

Unknown Host 25:41 cryptography.

Shweta Agrawal 25:42 Yeah.

Unknown Host 25:42 Is this related?

Shweta Agrawal 25:43 This is related. This can be seen as a special case of functional cryptography. So here you're not actually computing on, uh, the private information, you're computing on sort of attributes of that information. So, uh, you know, think of a file, uh, whose contents are secret, but, uh, I encrypt it against some kind of access control policy. So the access control policy could say that, you know, only, uh, members of the computer science department at IIT Madras, you know, who, who are, uh, uh, staff or faculty, but not students, should be able to open this file. Something like that. Then, uh, I can give out keys to people, right, based on their attributes. So if you're a professor, you have, uh, the attribute professor. If you, if you're a student, you have the attribute student in your key, and it will kind of come together and work that, you know, only if your attributes satisfy the access control policy can the file be decrypted.

Unknown Host 26:43 Nice. Interesting. So, okay, I have a sense of this. Um, I also read something called lattice-based

Unknown Host 26:49 cryptography.

Shweta Agrawal 26:50 Yeah.

Unknown Host 26:50 Is this also a field of-

Shweta Agrawal 26:51 No. So this is, this is not, uh, uh, an application. Lattice-based cryptography basically means, you know, building cryptography from hard problems on lattices. So these- this is a class of hardness assumptions. This is a class of hard problems. So sort of these are problems that would replace, for instance, factoring or discrete logarithm or other, uh, problems that were quantum broken. And the cool thing about lattices is that, uh, maybe they receive a lot of attention now because, uh, uh, they are conjectured to be post-quantum secure. But, uh, the cool thing is that the mathematical structure that they offer is so rich and so different from the previous number theoretic, uh, structures that we had, that you can do all kinds of, you know, very cool new cryptography that was not possible, uh, before. So it's, it's not just a solution to... or, you know, a conjectured solution to the quantum threat, but it's, it's also a door of opportunity. Like, it, it can lead you to cryptography that- that's really new and different and much more powerful than what we knew before.

Unknown Host 28:02 Okay, cool. Um, I, I also want to, sort of from my reference point of how I understand, uh, the, uh, the word hard problem. Uh,

Unknown Host 28:10 in, in my engineering course, we, we learned the Navier-Stokes equation, and we heard that that's a hard problem. Is, is it the same thing?

Shweta Agrawal 28:19 So, I, I don't know about this particular hard problem, but essentially when we say that the problem is hard to solve, it just means that, you know, if I'm given the input... So let's take factoring, because I think everybody knows this problem.

Unknown Host 28:31 Sure.

Shweta Agrawal 28:31 So the factoring problem is that, you know, I give you a number which is the product of two large prime numbers, and I want to figure out the factor of that- of the factors of this number, right? So now a trivial way to solve this problem is kind of just go-

Unknown Host 28:47 Brute force

Shweta Agrawal 28:47 ... brute force, right? You know, try two, three, five, seven. You know, just keep trying all the prime numbers till you hit a factor. Now, this process, right, it takes exponential time in the size of the input. So if the number is n, right, it can be represented in log n bits, and this brute force algorithm will take time, uh, root n, which is exponential in log n, right? So, uh, basically, the, the conjecture there is that more or less, this is the best you can do. So I say more or less, because there are more clever things that you can do than just this-

Unknown Host 29:25 Brute force

Shweta Agrawal 29:27 ... brute force, but nothing that kind of changes the order of, of hardness significantly-

Unknown Host 29:33 Sure

Shweta Agrawal 29:34 ... for a classical algorithm.

Unknown Host 29:36 Okay, understood. I will go back to my notes and check, uh, whether Navier-Stokes qualifies.

Shweta Agrawal 29:40 [chuckles]

Unknown Host 29:41 Mm. Can I just zoom back a bit? And you are a professor in the computer science department.

Shweta Agrawal 29:47 Yeah.

Unknown Host 29:48 Um, the way, um, uh, at an undergrad level or at a postgrad level, people understand the word computer science engineering is quite different from how a normal person understands

Unknown Host 29:59 it. Because a normal person would think, "Oh, he's a computer science student, he does coding, uh, probably codes for some company." But that's not how, um, a computer science faculty would look at it, right? So I think you were explaining to me before this call that broadly, computer science has certain subfields. Can you just say that?

Shweta Agrawal 30:17 Yeah. So sure. So actually, uh, it's interesting what you said because, uh, yesterday I, I taught my first class in Theory of Computation, which is the course I'm teaching this semester, and I spent a good part of the lecture exactly debunking this myth. That, you know, uh, computer science is not-... about coding, uh, you know, not about programming. So there's a very famous quote by, uh, a computer scientist called, uh, Dijkstra, and he says that, "Calling our field computer science is like calling surgery knife science." So a computer is, is just a physical realization of, you know, some of the ideas in the field. But our field is not about computers, it's about computing. You know, what can be computed? What is a hard problem? What is an easy problem? Are there some problems that are as hard as each other, right? Can I have efficient algorithms to solve these problems? When I say efficient, you know, how much time does it take? How much space does it take? How much parallel time does it take? If these problems, uh, cannot be solved, can I build mathematical codes? So not program- uh, programming codes, but-

Unknown Host 31:29 Algorithms.

Shweta Agrawal 31:30 Algorithms. Can I build cryptography from it, right? So these are the kinds of questions that, uh, are asked in, in the science of computing. Now, the science of computing has developed a lot, right? So, uh, there are also a, a lot of really cool things that have happened since we started asking these questions, like we built computers, which is obviously very cool. We can now have fields that study, you know, how to build an operating system. You know, how to build compilers, how to design programming languages. So this is sort of the broad area of systems, right? And then, of course, there is this whole big data revolution with machine learning and artificial intelligence, right? So these are things, you know, really beautiful questions, right? Like, can machines think? What, what can a machine do vis-a-vis what a human can do? Can a machine be creative, right? Like, uh, can I, uh, train a machine to have a conversation with me? So these are some really cool things that are happening in the world of machine learning these days. So these are sort of... There's a broad spectrum of things that you could, uh, study within computer science, and programming is only one sort of small tool that gets used by many people.

Unknown Host 32:41 Sure. So, um, very crude analogy would be that, um, in mechanical engineering, the people who build engines-

Shweta Agrawal 32:48 Yeah

Unknown Host 32:48 ... are different from the people who drive the engines, because it's not the same thing.

Shweta Agrawal 32:52 Yes.

Unknown Host 32:52 But I can understand that the engineers who are building the engines need to know how to drive it, and would be good at driving it, because that's t- those are the people who have built it.

Shweta Agrawal 33:01 So I think the field is broad enough that this is not necessarily true. So in, uh... What I mean is, so for instance, I work in theory, right, and I work in cryptography. Now, it's not necessary that I know how to do programming in a certain programming language for a certain... You know, for building a compiler, let's say. So I have a working knowledge of the whole field. I have an in-depth knowledge of, you know, some part of the field, and it is not that I can do everything, even in terms of using. Like, uh, I, I cannot necessarily code in every programming language, uh, for example.

Unknown Host 33:37 Okay, so there's theoretical computer science, which is about algorithms and cryptography, and I guess it's a lot-

Shweta Agrawal 33:42 Complexity, like hardness of problems.

Unknown Host 33:44 Okay, hardness, complexity.

Shweta Agrawal 33:45 Yeah.

Unknown Host 33:45 And then there is, uh, systems, which is-

Shweta Agrawal 33:47 Systems, yeah

Unknown Host 33:48 ... operating system, compilers-

Shweta Agrawal 33:49 Networks, you know, computer networks, programming-

Unknown Host 33:52 How to realize them.

Shweta Agrawal 33:53 Yeah, they-

Unknown Host 33:54 Sure.

Shweta Agrawal 33:54 Yeah.

Unknown Host 33:54 Okay.

Shweta Agrawal 33:55 And machine learning and so on, which is, you know, a third big field, sub-field.

Unknown Host 34:01 Of course, the field of machine learning has exploded in the last, uh-

Shweta Agrawal 34:04 That's true

Unknown Host 34:04 ... maybe 10 years.

Shweta Agrawal 34:05 Yeah, yeah.

Unknown Host 34:06 I remember we, we spoke to Professor Ravindran, um, a few weeks back, a few months back, and he was saying that his first course in reinforcement learning had six students, uh, 20 years back, and now it has, like-

Shweta Agrawal 34:17 600. [chuckles]

Unknown Host 34:18 Yeah. Plus, he takes the online course, which has, like-

Shweta Agrawal 34:21 Yeah

Unknown Host 34:21 ... 6,000.

Shweta Agrawal 34:22 Yeah.

Unknown Host 34:23 So when people think of, uh, joining computer science or at a undergrad or at a post-grad level, um, why should people enter this field?

Unknown Host 34:31 What are the exciting things that are happening, and what should their proclivities be? Like, how will I know I'm- I'll say I'm interested in it. First of all, why should I be interested? And if I'm interested in it, what should I already be good at to excel in the field?

Shweta Agrawal 34:46 Right. So I think that, uh, the answer, of course, depends on what level you're talking about, so let's start with undergrad. I mean, so I think, you know, in IIT, it's easy to know if you're good at it, because even to get admitted, you have to, uh-

Unknown Host 35:00 Cross the-

Shweta Agrawal 35:01 ... cross the joint entrance exam. So you will pretty much only get here if you're good at it. But, uh, why should I be interested in it is actually an excellent question, and it's something that I think that, uh, students and, and their families don't ask enough. Because maybe we're still kind of in this mentality, by and large, where, you know, we're just thinking of what school, you know, or what'll get us a high-paying job, and, and, uh, things of this kind. Whereas, you know, maybe you could be, uh, a very good writer or, you know, a very good, um, biologist, right, or, uh, a very good architect. So I mean, there are lots and lots of, uh, fields of study, like history, right, uh, psychology. I mean, each field is very rich in its own way, and, uh, what you're interested in ideally should, uh, get determined by your very individual nature and, you know, uh, a, a good exposure to a wide variety of different topics. And ideally, like, there should be, like, an informed choice that is made. Of course, this does not happen, but, uh-

Unknown Host 36:10 Sure

Shweta Agrawal 36:10 ... that's the ideal scenario. And, uh, in terms of why should I care, I mean, if you're good at it, first of all, I think that's good enough reason to care. But, uh, also additionally, you know, it's a great place to be. I mean, computer science is a very young field. There are lots of deep questions. Many of them are still open, right? And like cryptography, the field in which I work, I just can't think of a more fun field to be in, because, you know, it... There's just so many really exciting questions, and-... I feel like so many big discoveries are happening now, and, you know, we can be part of that. So it's, it's just a great time. So I think that whether undergrad or post-grad, you know, this kind of thought process, uh, has to be there. Uh, at the post-grad level, of course, it's much more, uh ... The hope is that, you know, it's much more informed. If, if you did your undergrad in whatever field was right for you, right? Then you've had an exposure to, uh, topics that might be potentially interesting to you, and then the post-grad is a place to kind of, uh, go deeper into something and kind of really get to business.

Unknown Host 37:19 Nice. So you're saying that at an undergrad level, of course, I mean, other than the fact that JEE sort of forces you to pick subjects, you should be aware of what your interest levels are?

Shweta Agrawal 37:29 Yeah. I mean, in India, we're still quite far from that, but ideally, yes. I mean, there are students, you know, we see who basically, uh, are in the institute and, you know ... For instance, I had a, a student, I was her faculty advisor, and she just did not like it at all. Like, she did not like the course, she did not like the subjects. Uh, she wanted to do art, like, she wanted to be a writer. But there was so much pressure

Shweta Agrawal 37:56 on her to kind of stick on and, you know, she was trying to get out of the program. And her, uh, her parents would call me and say, "Ma'am, please don't let her quit," like, "You're her faculty advisor." And I would be like, "You know, i- this is not fair. Like, you need- she's an adult, and she needs to make her, her decisions." So-

Unknown Host 38:16 Would it be fair to say that computer science students, in particular, are under a lot of pressure, uh, because they've got here, [laughs] and then, you know, there's this, like, their expectation from-

Shweta Agrawal 38:27 Pressure cooker.

Unknown Host 38:27 Yeah.

Shweta Agrawal 38:28 Yeah. It, it is hard. I think it ... Yeah, it, it is, uh-

Unknown Host 38:31 Different kind of hard problem. [laughs]

Shweta Agrawal 38:32 [laughs] It's a different kind of hard problem.

Unknown Host 38:35 How is, uh-

Shweta Agrawal 38:35 But one that you want to solve.

Unknown Host 38:37 I know. I agree. Uh, and, um, maybe with the one thing that I do like about the IIT Madras system now is that, uh, at the very least, there are electives students can, uh, if they want, and if they allow themselves, can explore, uh, beyond their chosen field.

Shweta Agrawal 38:54 Yeah, for sure you can explore. I mean, uh, having said that, in India, we still, uh, don't have that many very strong universities, you know, where you can really explore an undergrad, not just in ... Like, I, I didn't come in just to do engineering, you know? I, I came in, and now I'm exploring.

Unknown Host 39:14 Yeah.

Shweta Agrawal 39:14 So that's a dream-

Unknown Host 39:16 Fair enough

Shweta Agrawal 39:16 ... that's a dream setting.

Unknown Host 39:18 The memory that comes-

Shweta Agrawal 39:19 We're still a technology, right, college, right? So we still-

Unknown Host 39:21 Memory that comes to my mind is that, uh, you know, there's one place where Steve Jobs talks about how he had explored a calligraphy class.

Shweta Agrawal 39:29 Yeah, and that, yeah-

Unknown Host 39:30 That him design the first fonts for Mac.

Shweta Agrawal 39:33 Yes, yes, yes.

Unknown Host 39:33 Uh, that's maybe quite- that, that kind of broad exposure-

Shweta Agrawal 39:39 Absolutely

Unknown Host 39:39 ... is quite limited.

Shweta Agrawal 39:40 Yeah.

Unknown Host 39:42 Can I ask you, in the field of cryptography, um, where is, uh, India placed in the world, uh, or IIT Madras placed in the world? And maybe you can give us some sort of historical context. How has it moved in the last 10 years?

Shweta Agrawal 39:53 Yeah, so 10 years is a good number that I can talk about, because I moved back to India about 10 years ago after my PhD and post-doc, and, uh, I think the field has come a very, very long way since then. So when I moved back, uh, there was almost nobody really, uh, working in public key cryptography. So there was, you know, ISI, which was doing some symmetric key crypto, but, like, in the, uh, public key space, there was almost nobody. And, uh, now in the last 10 years, actually, we have a, a very strong body of, uh, cryptographers. So for instance, you know, our top, uh, global venues for publishing kind of new, new research discoveries, right? Uh, these are like crypto and neurocrypt. And, uh, when I moved back, there was almost no representation that India had in, in these, uh, venues. And now if you look at it, actually, you know, every year at, at these top venues, we have a pretty good representation from India. So more people have not only come back, more strong people, but also they have trained their own students, right? So now we're actually generating strong cryptographers who are in turn contributing to the field from within the country. So it's, it's a really very positive movement. And at the time that I came back, you know, it, it was, it was just an idealism and a kind of a dream, but I, I really feel grateful that, uh, in fact, it, it, it has, uh, it ha- it has been so good. Like, I, I did not really know what to expect. I just knew I wanted to contribute to this. But in fact, it, it's been so fantastic, I r- I really feel very happy and proud that-

Unknown Host 41:31 I want to inquire more about the why you came back, but before that, right now you are leading a center called CyStar?

Shweta Agrawal 41:38 Yeah, I'm co-leading it.

Unknown Host 41:39 Okay.

Shweta Agrawal 41:39 So there are three of us, uh, that lead different aspects of it. I lead the theory aspect. So CyStar, uh, there's a joke about the name, which I'll tell you in a second, but, uh, CyStar is, you know, a, a center for, uh, security- cryptography, security, trust, and reliability. That's the name.

Unknown Host 41:57 Okay.

Shweta Agrawal 41:57 Uh, the internal joke is that I was pushing for this name because, you know, psi star, like, uh, psi star is the conjugate of the wave function. So psi is a Greek symbol. Anyway, this is an internal quantum joke, which I did not tell anybody, but this is why I was pushing for this name, and, uh, I'm very happy that everyone agreed to this name. So yeah, but, uh, coming back to CyStar, geeky jokes apart, [laughs] uh, it's, uh, it's a center of excellence in cybersecurity. And, uh,

Shweta Agrawal 42:29 you know, we're, we're just very, very happy and excited about it, because it's, uh, something that's finally kind of letting us come together and do more than a sum of the parts, you know? So, uh, there's three of us that- whose vision it, it's, uh, sort of, uh, driving, right?... uh, it started as like a small research project that, uh, we applied for, my colleague John and I. And then, uh, it's grown in sort of phase two of the funding, our, uh, sort of third partner, Chester, joined. There's also other affiliated faculty, of course. But the three of us look at different aspects of cybersecurity. So Chester looks at, you know, very applied, like implementation, you know, leakage attacks, so the applied side, which is directly society-facing. And, you know, uh, unsurprisingly, I think like, uh, that gets kind of the most, uh, excitement gets generated around this because-

Unknown Host 43:24 Because the, the, it's a near-term-

Shweta Agrawal 43:26 Yeah

Unknown Host 43:26 ... impact.

Shweta Agrawal 43:27 It's near-term impact, and it's also more relatable. And then I work, of course, on the algorithms, and that's also the, the vertical that I lead, so this is, uh, we've already talked about. And then John is, uh, you know, the bridge, so he leads this, you know, a kind of defining, uh, a theoretical models coming from the real world or applying theoretical ideas to real-world problems. So he's really like a distributed computing guy who understands aspects of both and makes connections between them. So we, we now have this... And, and we've, uh, hired, you know, s- uh, some amazing staff, like administrative staff, which helps us to do a lot of outreach, uh, you know, hackathons, winter schools, you know, startup incubators. And, uh, it helps us to pay our PhD students more. You know, we're hoping it will help us to attract more students. Uh, so it's, it's just a way to sort of really, um, dream big and, and, uh, put those dreams-

Unknown Host 44:31 Interesting

Shweta Agrawal 44:31 ... on paper.

Unknown Host 44:32 The, the other centers that are there in, in IITM are sort of like, um, more, um... I, I don't know if this is true, but you can correct me, are interdisciplinary, have a broader focus. Here, it's a clear, focused effort around, uh, cryptography, right?

Shweta Agrawal 44:47 Not cryptography, cybersecurity.

Unknown Host 44:49 Cybersecurity.

Shweta Agrawal 44:49 So cryptography is just the map of cybersecurity, right? There's a lot more. I mean, there's systems building, right? So there's systems, there's applications, so that's actually the point of the center. And cybersecurity is so wide that already, like, we cannot do all the things that cybersecurity is. So for example, you know, we don't have people whose expertise is, you know, uh, machine learning for cybersecurity. So there- I'm trying to say it's a very wide field, right? Theory is, is one, uh, aspect of it. There's a lot more. And, uh, as a center, what we're doing is, you know, we're, we're trying to play to our strengths, so we don't claim to do everything. But we're trying to take a holistic view of the problem of cybersecurity, bring in all these different perspectives, and try to implement, uh, a vision, like I said, which sort of spreads the excitement, the awareness, you know, tries to really build experts for the country.

Unknown Host 45:49 Nice. Since you shared the, the internal story behind CyStar, I want to share something with you.

Shweta Agrawal 45:56 Yeah.

Unknown Host 45:56 Uh, when CFI was set up, um, maybe about 15 years back, I was involved in the naming.

Shweta Agrawal 46:02 Oh.

Unknown Host 46:02 And, uh, CFI is called like that also because the number phi, the Greek letter phi, is used, uh, as a letter for phase, and C is the letter used for constant.

Shweta Agrawal 46:16 Oh. [chuckles]

Unknown Host 46:16 So it's constant phase, and the assumption is that it's a lead, uh, it's a leading angle, so it's a constant phase lead, so C phi. Uh, but we needed to sort of integrate it with-

Shweta Agrawal 46:27 Yeah

Unknown Host 46:27 ... everybody understands, so CFI.

Shweta Agrawal 46:29 So similar to CyStar.

Unknown Host 46:30 Yeah.

Shweta Agrawal 46:30 [laughs]

Unknown Host 46:31 Nice. Professor, we spoke a lot about the work you're doing and the field itself. Um, but also I have seen you, uh, in various forums

Unknown Host 46:40 talk about women in STEM. Would you like to sort of share your opinion and your views on that?

Shweta Agrawal 46:45 So I mean, is there a particular question that you have?

Unknown Host 46:48 I don't know. I feel like sometimes I'm so removed from this topic, uh, also because technically I'm, I'm a marketing person, not even in STEM, right? Um, I'm not even aware of the challenges. I broadly know that there are fewer women in STEM than men.

Shweta Agrawal 47:04 Yeah, I think actually in India, the numbers are not as, uh, skewed as, uh, even in some developed countries. I think the challenges in India are quite different than, for instance, in the US, where I, uh, did a lot of my education. Uh, I think in India, the challenge is not that at the undergrad level people think that women cannot, uh, you know, go into STEM. Uh, in fact, we have lots of girl students. Maybe not at the IITs. That's a different, uh, problem. I mean, that's because the IITs are not co-located with homes, right? There, there are few in the country, and, uh, students have to travel to-

Unknown Host 47:44 It's a residential.

Shweta Agrawal 47:45 It's a residential college. So yeah, parents of girl students, uh, are often much more reluctant to send, uh, their daughters. Of course, this is changing quite fast, but, uh, overall-

Unknown Host 47:58 Not just the IITs are residential, but the, the... maybe some of the coaching institutions and-

Shweta Agrawal 48:03 Yeah, right, right

Unknown Host 48:04 ... the-

Shweta Agrawal 48:04 Leading up to it, yeah.

Unknown Host 48:05 They're also residential-

Shweta Agrawal 48:06 Yeah

Unknown Host 48:06 ... they're also, are difficult.

Shweta Agrawal 48:07 Right, right. So, uh, y- yeah, but I think that, you know, in un- at the undergrad level in general, we do see a lot of girl students. The problem is more that they sort of, uh, start... This, this crowd thins more and more as you go up, and this is due to kind of various reasons, you know, um, many of them are social, cultural, you know, perception issues, uh, and so on. So, uh, there are certainly challenges for women in STEM, and, uh, this is well documented. Like, so there is, for example, you know, this, uh, uh, famous experiment which was conducted several years ago called the head of the table experiment. And in this, uh, experiment, like-... there's a picture, there are two pictures where, uh, there's a table, and there are a bunch of people sitting around this table.

Shweta Agrawal 49:02 And, uh, there's one chair which is clearly the head chair, and, uh, in one version of the picture, there's a guy sitting in that, and in one version of the picture, there's a girl sitting in that. And then, uh, these pictures are shown- one of these two pictures, uh, is, is shown to, you know, random people, and, uh, they are asked, "Who do you think is leading this discussion?" So when there's a guy in the head chair, right, 100 out of 100 everyone points to him. When there's a woman, 50 out of 100, they point to the guy next to her. So this is actually a very old study, but I would not say that this has changed very much. I think that particularly seeing women in leadership positions is something that society still grapples with. And, you know, you face resistance in un- unexpected quarters. Because I think that even though now more and more women are accepted as part of the workforce, right, um, I think the litmus test is really disagreement. Like, there is kind of a template of... See, this is all, you know, very subjective, but this is my, uh, perception of, of the issue.

Unknown Host 50:12 Can you give me a minute?

Shweta Agrawal 50:13 Yeah.

Unknown Host 50:14 Let me just, let me just rephra- let me just sort of mirror what you have said so far, uh, and understand it right. So if I understood right, what you're saying is that at a school/college level, there's not much difference, uh, in, say, in science grades, then-

Shweta Agrawal 50:28 No, no, grades actually, you know, women do very well in the grades. The point is whether they continue-

Unknown Host 50:34 Continue.

Shweta Agrawal 50:35 Yeah, yeah.

Unknown Host 50:35 Right, so it's-

Shweta Agrawal 50:36 So did you know that in the IITs, the women students have a full point higher CGPA on average?

Unknown Host 50:42 Is it?

Shweta Agrawal 50:43 Yes.

Unknown Host 50:43 Okay.

Shweta Agrawal 50:43 This is a study done in IIT Delhi.

Unknown Host 50:46 Okay. So at a, at a IIT level, maybe there are about 20, 20 to 30% women, and then as you grow higher in STEM, it sort of thins down.

Shweta Agrawal 50:55 Yeah.

Unknown Host 50:55 At, at a leadership level, it's very less.

Shweta Agrawal 50:57 Very, very less, yeah.

Unknown Host 50:59 And Professor Preeti, of course, now is recognized as-

Shweta Agrawal 51:01 Yes

Unknown Host 51:01 ... IIT's first IIT woman director.

Shweta Agrawal 51:03 Yes. Yeah.

Unknown Host 51:04 And that-

Shweta Agrawal 51:04 And I'm really proud of this, and Preeti is amazing, and, you know, I'm really happy that, uh, she's doing this. And also, you know, that, uh, we're doing this, that IITM did this. It's really amazing.

Unknown Host 51:16 Yeah. Hopefully we'll get to interview her also [chuckles] soon, when she is here, but she's mostly in Zanzibar now.

Shweta Agrawal 51:21 Yeah, right.

Unknown Host 51:22 Um, the other thing that, and just to sort of lead into the last segment of this interview, um, at some point, uh, we spoke about it, but let's come back to it. Um, your

Unknown Host 51:35 own personal journey, um, leading to a decision of, uh, your returning from US, uh, can we inquire a little bit about that? Maybe you could start with your undergrad or your schooling days, or, you know, your-

Shweta Agrawal 51:47 Well, so I feel like my whole life journey, uh, is really different from how I see undergrads now, in the sense that when I see many students now, uh, they have very clear ideas of, you know, what they want to do. Like, everything will be very well-researched, you know, where to join, what to do, you know, next steps, all the pros and cons will be weighed, and so on. Uh, in my particular case, I think that I never used to take anything super seriously, and I was just doing more of a random walk, like, uh, you know, just trying to explore things that I found interesting. In growing up, I felt that, um, there was a lot of struggle and strife that I saw around me. Like, I came from a middle-class background, and opportunities were pretty few and far between. Like, when I see, again, you know, my, uh... I see children of our colleagues here, right? So they have so much opportunity. But back in our generation, I mean, you know, we, we didn't really know much. We weren't exposed to much. And, uh, I just happened to do, you know, computer science because of... That was the pressure that, uh, you know, society placed on us. And, uh, there were various other things that I liked. I, I particularly loved art, and I was also very good at it, and, um, for a long time later, I felt like I should have been an artist. But luckily, I became a mathematician, and it's the same thing, so, you know, I'm, I'm happy where I am now. But, uh, yeah, I did my undergrad, and I didn't have really any exposure to research or anything. And at the time, I would be reading these, uh, popular science books, you know, by Stephen Hawking and, and so on. And there was this one particular book which really made an impact on me, and this was by Brian Greene, um, called The Fabric of the Cosmos. And, uh, I think back then, I mean, I was already in this, uh... I, I used to feel like, you know, s- something exciting is happening somewhere, and I just need to find it. And when I read this, uh, book, it was just so mind-blowing. Like, you know, there's a chapter where he explains general relativity in a way that, you know, an amateur can really understand. And, uh, I think that in some sense, that really, um, catalyzed my desire to find something, you know, uh, new and exciting. And, uh, I just did, like, a brute force application to a whole bunch of universities. Uh, I got selected, you know, I went abroad. Even when I did that for my master's and subsequently PhD, I was still exploring a lot of things. In fact, I took a lot of art classes. I took a lot of math classes. I took some physics classes. Of course, I took computer science classes. So I was just very excited to learn.

Unknown Host 54:42 On your post-grad in the US?

Shweta Agrawal 54:44 Yeah, this is during my master's and, uh, and PhD. So cryptography also happened, you know, really by chance. Like, I was visiting a professor. It was supposed to just be a short-... one or two-week visit about some particular question, and he was a cryptographer. And, you know, uh, during that time I got exposed to cryptography. I just started exploring it, and eventually, I mean, you know, that, that's the field where I, uh, really committed myself. So, ah, this is how my, my education was.

Unknown Host 55:19 If, if the field of cryptography is, uh, at the time that you're talking about, maybe in the mid-2000, mid-2000s, early-

Shweta Agrawal 55:25 '10. Yeah, 2010, let's say.

Unknown Host 55:27 If the, if the field of cryptography is more evolved in the US and, uh, wasn't so evolved here, um, what led for you to take the decision to come back?

Shweta Agrawal 55:38 So, uh, this, this was just a personal decision. Before I even left, I knew I would come back, and I was in fact 100% clear about it. Uh, and, uh, this is because, uh, as I said, you know, while growing up, uh, I had very little exposure, and I saw a lot of struggle around me, and, uh, uh, I really felt for it. And eventually I started feeling that, uh, you know, somehow in this developing country where opportunities are often so hard to come by-

Unknown Host 56:14 Yeah

Shweta Agrawal 56:14 ... so scarce, you know, so many talented people don't find these opportunities. I felt that I was somehow profoundly fortunate to, to have received the best of what this environment could offer.

Unknown Host 56:26 And you used the word random walk, which by the way, is also a name of algorithm, right?

Shweta Agrawal 56:30 Yeah. I mean, it's, it's... Yeah, it's, it's, uh, not one algorithm, but, like, a class of-

Unknown Host 56:35 Okay

Shweta Agrawal 56:35 ... algorithms. But, uh, yeah, so I, I really, uh, cared about coming back to this environment and giving back whatever I could. So even before I left, it was clear to me that I would always come back, and as soon as I finished my education... Uh, during my education, I also met my husband and, my now husband, and, uh, he had the, the same dream. And so as soon as we finished our education, we j- we just came back, so we did not even apply for jobs there. And I, I mean, uh, I'm very happy with this decision. No regrets.

Unknown Host 57:11 And we are lucky for it. [chuckles]

Shweta Agrawal 57:13 Yes. [laughing]

Unknown Host 57:15 I think you m- you mentioned, uh, maybe, uh, we, we can probably stop here, but we'll just explore it and see if it works, and then we- otherwise we'll cut it off. Um, Professor, uh, during our preparation, you also mentioned, uh... And I, I just want to note that, uh, to the audience, that the preparation for this interview was the hardest, because I know nothing about cryptography, and my research team also knows nothing, so we actually ended up learning a lot. Professor, during our preparation, you also used the word Atma Shraddha. I, I love the way you explained

Unknown Host 57:43 it. Um, could you do it for our audience?

Shweta Agrawal 57:46 Well, it was in the context of something we were discussing. I, I feel that, uh, you know, we have everything that it takes. Uh, like, we as a country, we as an institute, we have everything that it takes. We have all the potential, but we're still not where we need to be, like, you know, in science, for instance, right? So, uh, I feel that this is because somehow we're still in this post-colonial mindset where we feel that we need to go out to be cool, like to be, you know, to, like, to be part of the excitement or, or, uh, something. And by the way, as I already, uh, said, even when I was a young person, you know, um, I, I did go abroad. At that time, uh, these opportunities were not there. Like, now, you know, things are significantly better. There are a lot of groups here that are on par with, like, you know, the, the best in the world. But still we, we see that students many times, you know, they just want to get out. This is not based on any fact or any, you know, objective measure of, you know, the scientific output or the funding opportunities, or even the salaries, which, you know, can be very good now. But it's just based on this perception that, you know, this place, this country is not g- good enough. I, I just need to get out.

Unknown Host 59:12 Sure, and, you know-

Shweta Agrawal 59:12 So this is... Yeah.

Unknown Host 59:14 Y- you're from a science field, but i- in my field, in marketing, I've seen people who have worked with us for one or two years, and they apply to MBA colleges abroad. And I'm looking at the college they have applied to, and I know it's not a great college, and I have zero idea why they're going.

Shweta Agrawal 59:28 Yeah, I feel the same way. Like, uh, for instance, I have a course on quantum algorithms on NPTEL, which is our, uh, in-house MOOC, as you know. And, uh, you know, there are students who see this course and who write in to me about it. And I won't name, uh, any specifics but, like, you know, recently there's a student sitting in, uh, you know, university X in country Y, and, uh, writing to me saying that, "You know, I'm trying to work on this topic, and I've been watching your lectures, and can you guide me in blah and blah?" And this student has done his undergrad at a very good college in India, right? And has gone to this university X in country Y, right, which is nowhere compared to our own group here or other groups in India. Like, I'm not talking only of IIT Madras or, you know, our particular CyStar or something, but I mean, we have some fantastic, uh, researchers in India, right? But students, they... I- this is where I was using the word Atma Shraddha. I feel that we lack this, uh... You know, as a people, we're still not fully there. We, we imagine that we don't have it by ourselves. Like, we, we need to get out, and only then we'll find what we want.

Unknown Host 1:00:48 And I want to share another story on this. Uh, one, one of the times, uh, as part of our work with Ask IITM, one of the times where we were doing a sort of a, a, a sort of a roadshow, you could say, uh, in Bangalore, and we were describing all this work that students do here, all the work that professors do, and we were talking about the Hyperloop project that happens here, and-... I think there was like a sort of a cognitive dissonance with the audience, because the audience was in the mindset that this kind of work happens only, you know-

Shweta Agrawal 1:01:16 Abroad.

Unknown Host 1:01:17 Abroad. Only Elon Musk-

Shweta Agrawal 1:01:18 This exactly what I mean by Atma Shraddha. You see, that, you know, you, you cannot, because you yourself relate to other Indians, right? And you don't have confidence or faith in yourself, you imagine that nobody here can be doing anything, uh, you know, really top class or important or, or whatever. So you don't even believe it.

Unknown Host 1:01:45 Yeah.

Shweta Agrawal 1:01:45 Like, if we tell students, right, that, ah, for example, look at objective measures of, uh, yeah, f- for instance, in research, right? There are measures like citations or whatever, like diff- for different fields it could be different. But they're not even at the place where they can start evaluating this question, because there, there is such a strong preconceived notion that I/we cannot do it, you know? Or at least cannot do it here. So maybe I can do it if I ship abroad and, you know, it doesn't matter even where. Like, so by the way, it's not that I'm saying you should always be here, right? You need to make an informed choice, like, what are your options, uh, you know, what, what are you good at? Like, what is the group good at, et cetera. So just an informed choice, and indeed, it could be the case that some institute abroad i- is better for you for your interests. That could certainly be the case, but it's not a given.

Unknown Host 1:02:44 Yeah, it's not universally true.

Shweta Agrawal 1:02:46 It's not universally.

Unknown Host 1:02:47 And, and-

Shweta Agrawal 1:02:47 Yeah

Unknown Host 1:02:48 ... for specific research areas, uh, we are actually doing really well.

Shweta Agrawal 1:02:52 We're doing very well, yeah.

Unknown Host 1:02:54 And it would be almost, like, silly, I, I'm using a mild word, silly for students to assume that they have to go abroad, and-

Shweta Agrawal 1:03:01 It's not just students, right? I mean, it, it's our larger population. Like, uh, we, we need to develop more confidence in ourselves, I think. Yeah, more Atma Shraddha. [chuckles]

Unknown Host 1:03:12 More Atma Shraddha, and hopefully, uh, through these podcasts, uh, we can establish that. This is really the Best Place To Build. Thank you so much. But before we leave, uh, I wanna ask you if you are, um, still doing

Unknown Host 1:03:23 art? I know you said that you do it as a student, uh, if that's something you do as an adult.

Shweta Agrawal 1:03:26 Absolutely.

Unknown Host 1:03:27 And, uh-

Shweta Agrawal 1:03:28 I do, I do do art, and, uh, uh, as I said, you know, cryptography itself is a kind of art. So in fact, I would say cryptography is a lot like abstract expressionism, which is one of my favorite movements in art. Because if you see abstract expressionist paintings, right, they are also a, a dance between structure and randomness, and that's exactly cryptography.

Unknown Host 1:03:50 Oh, that's beautiful.

Shweta Agrawal 1:03:52 [chuckles]

Unknown Host 1:03:52 Because it's like I'm trying to tell you something, but not everybody will get it.

Shweta Agrawal 1:03:55 Ah, no, no, no, it's actually... It's actually, you know, this, uh, this sort of, uh, dialogue between form and formlessness, you know? It, it's there, but it's not quite there. So I mean, of course, the cryptography is, is, uh, one example of this in, in my mind, but of course, like in the artistic movement, there can be many ways of expressing this idea. Cryptography, in my mind, is one of the ways. And, uh, more broadly, I think that these distinctions between fields are all, um, you know, unnecessary. Uh, like, what drew me to art is, is a love for, uh, a, a sort of pursuit of beauty, right? And, and this is also something that drives me in, in math or, you know, uh, helps me to appreciate music or poetry or history, I mean, psychology. So I mean, it's everywhere, like, uh, beauty is everywhere, and, uh, you just have to approach it.

Unknown Host 1:04:57 I think that's a great place to stop. Thank you for

Unknown Host 1:05:00 joining us. Um, if you like this podcast, please like, share, and subscribe, uh, and, uh, follow, uh, Professor Shweta. Uh, if possible, view her NPTEL lectures, and, and join her in her courses. Thank you.

Shweta Agrawal 1:05:14 Thank you so much. [upbeat music]