How to Survive the AI Wave as an Engineer (IIT Kharagpur Grad POV) - w/ Aayush

Everybody says that you have to leave India to get a master's degree that is actually useful. Aayush had the GRE scores. He could have easily gone to the US or any European country. And yet he chose to get his masters from IIT Karakpur. Karakpur. Let me go ahead in my country and at a very economical rate and proximity to my house. I'll be able to build a strong launch pad. Not only that, Aayush gained deep theoretical expertise in the domain of machine learning, got an amazing placement and also built a blueprint for every young engineer in India to ride the AI wave without leaving.

the AI wave without leaving. Tier 2 and tier three students are very talented students. What they really need is some kind of platform. We talk about the money, compare a masters in India versus one abroad, the true reality of an Mtech degree from an IIT, the exact path to mastering AI without leaving India and why India lags behind in development of foundational AI models and also a fascinating insight on how humans can learn from AI to actually improve our lives. Our society has a subjective bias if you believe it or not that if you are from an IIT you are seen differently and if you are from a nonIT let's say we have a student in ocean engineering or civil [singing] architecture or civil engineering they are very fascinated about AI so how do they approach it this

about AI so how do they approach it this is the ready set do podcast where we focus on just the first few steps at any endeavor and without any further ado let's dive in welcome to the only podcast in the world featuring stories of high agency individuals who are just a few steps ahead of us. Aayush, welcome. Yeah, thanks for having me. Nan, ever since we calendared this, you know, I've been just uh obviously researching, brainstorming, just things that I want to cover with you. Where I want to start here is given you your you are currently in your master's curriculum at IIT Karakpur. Um there are a lot of myths or misconceptions at least in India around who a technical masters is for and who it is not for. So I'm hoping that

it is not for. So I'm hoping that because you've been in that journey right say there is a young graduate a bachelor's in you know computer science or electronics or what have you any engineering field say they are nearing the end of their degree and they are faced with the option that probably you were also faced with even though I know that you did you did get some work before you joined the masters but essentially my question is who is like an Indian masters for and who is it not for basically So the master system in India I want to answer to this question of yours uh regarding the educational institution of the bachelor's program. So in India we because we have to first of all get the

because we have to first of all get the flow like how people think. So we have students coming from tier 1 colleges where we have the IITs and some good state engineering colleges and then we have tier two and tier three colleges. Now most students who like really want to go for a who think about masters are those from tier 2 tier three colleges and what is the fundamental thought because firstly in our country uh the first aspect that students think about is the placements correct and in most countries and yeah but given the fact that uh I have been drawn and brought up here so my question will focus more on the Indian context. Okay. For sure. So what many students think is that they want a good placement. Now in India what

want a good placement. Now in India what happens is that most of the youth they do not they don't want to join uh companies at a very low CTC because they feel that they are young and they want to learn they want to invest the best years of their lives into something more productive into something more interesting and which has more rewarding outcome. So when we think about students from tier 2 and tier three colleges for them mast's program is like a very important thing to do because most of them know that for them getting a placement getting a good placement is kind of going to be a very herculan task. Yeah because of the campusing condition in tier 2 tier three colleges in our country. So that is the first motivating factor that those individuals think of.

factor that those individuals think of. There are some set of individuals who who are placementoriented and who get a good placement from tier 2 tier three colleges and maybe they don't think about masters but here first of all we are thinking from a placement perspective perspective correct correct I will cover the learning perspective later on but from a placement perspective a tier 2 tier three student always feels that no if I get a mast's from an IIT or an IA or maybe a top nit then my scope will be more broad and the reason behind it is that it's uh not just about getting a good job but then once you go into the industry because I have been in an industry and I

because I have been in an industry and I have looked at the people who were director and technical fellows experts in their fields and all of them happen to have a higher degree. So a higher degree see not only just kind of gives you the door to a good job you the gives you the door to a good job with a good learning opportunity but it also gives you the potential to have more faster learning and when you do a masters from a good institution in an India not only it's about placements but it also levels up your uh network your connections because you maybe get to connect with the alumni so it builds a very strong foundational base among the students uh who do that. So placement of course it's a positive aspect for the student. So that is one aspect like why Mtech is for now uh let's talk think about learnability like in my case

about learnability like in my case I was just going to ask you yeah so I did my Vtech from a tier one college a good college and then I went into industry but then eventually I felt that I want I was initially I was actually inclined towards the mathematics side and I really wanted to understand the this core fundamentals behind AI and ML because when I was in my fourth year of my undergrad. I had taken up machine learning courses by professor Andrew Nji. I had gone through uh his lectures in the Stanford online and that kind of really propelled me and uh I had this kind of a subjective thought in my mind that maybe AI is the future. Maybe there is a lot of excitement not just hype but also from a

excitement not just hype but also from a mathematical aspect because AI takes a lot of good mathematical ideas from so many domains. So there are students like me as well who are interested in the learning aspect. So who who think that okay like if they are from tier one and they have some kind of placement because even in JU I had seen my colleagues uh taking up mast's program and even though they had an offer especially in India.

they had an offer especially in India. Yeah. And that is the reason because they want to have a explore more options. They want to even broaden their horizon because uh yeah because just a bachelor's degree is okay. It's good and but adding a master's degree you know it it it kind of changed the like it can be a completely different ball game for sure because the because when but I uh kind of you know uh came to prior to coming to IIT Karakpur and uh at that point of time even I was not very much active with LinkedIn and I came for my master's program I shared the journey and I was able to grow 10x uh my community by by 10x. So that is one benefit you get. But

10x. So that is one benefit you get. But let's talk about the other aspect that you mentioned who it is not for that is again important because you mentioned for whom it is poor and for whom it shouldn't correct right nam if you really have some decent job offers at hand. All right. And if you feel that you know if I if you just only think about the placement perspective and you already have like let's say a decent placement after your BTE. Yeah. Then maybe some students or maybe a good fraction of them might not consider a master's program because what happens is that to get into an Mtech you have to give two you have to give rigorous graduate aptitude test examinations. Right.

aptitude test examinations. Right. Right. you know you need to get a good rank in that then then you need to you know pass the interviews then you need to stay here for 2 years you need to invest in two years and uh there is a cost as well there is a monetary cost as well that you have to sacrifice so I think it becomes very subjective as well for who wants to pursue and who does not want to but objectively speaking somebody with a good placement would not that much of tempted to pursue a master's program. But again, as I said, it's a it's a it's it's a subjective choice and there are some people who went with their masters and

people who went with their masters and they completely transformed themselves when they end of graduate from the programs. So I think that kind of fundamentally uh I hope maybe answers the your question to some. Oh, sure. No, no, it it completely answers it. And you know just in summary cuz you know that was such an interesting and detailed deep dive. So you said that in summary people that don't have the or that haven't gotten the best placements they should consider masters. People that are looking to learn or get into research should also consider masters. And then also people that are from like tier two or three colleges that don't have the greatest networks that probably are looking to level up their career just in terms of credentials. they should looking at the and here if I kind of interrupt you in a

and here if I kind of interrupt you in a bit I do feel that tier 2 and tier three students are very talented students what but what they really need is some kind of platform right to kind of wait on their resume to like wait on their resume and so that people take them seriously because what I have seen is that there are a lot of people who have a lot of talent who work very hard but just because they are not getting that attention because they are not from that college because our society has a subjective bias if you believe it or not that if you are from an IAT you are seen differently and if you are from a nonIT you are seen differently

differently around you the moment you feel into the the two eyes three eyes four eyes whatever you have it is but it is yes but it is obviously exists and I strongly believe in it that a very strong student in tier 2 tier three college who has lot more knowledge can flourish and grow himself much better than an Ian if they get into that deck program because that kind of totally transforms them you know and uh so that was one aspect of it it's it's just a kind of a gray area right but it's so interesting for me to think about that like say there is in fact a tier two or three student say they get a job it's not the greatest job ever it doesn't pay the most money ever

ever it doesn't pay the most money ever but it is something. So tell me about this track where they just take the job and they like grind really hard. They do their upskilling on the side and then they are already starting day one already looking to switch you know switch to better offers better companies and then bounce around until they get to that same pay grade that they would have gotten to even after their master's degree and having spent 2 years and a few thousand well it's more than a few thousand rupees but you get what I mean what you were mentioning about the monetary monetary so yeah that is a very good question and I think that applies to me as well because the kind of question that you have posed on to is that because switching seems to be a very you know a a very kind of lucrative option that would only have to

that would only have to outside yeah I don't know anything about that but I am just yeah that but that's the premise of this question yeah think about it that when I'm trying to build a building when I have a vision to build a big monumental building so the foundations of the building how tall the building will grow is determined by the foundations of that building, the roots of that building. Yep. Yep. A good Mtech program offers you very strong fundamentals, very strong foundations. Because to build strong foundations, what you really need? See, industry is about more about implementation than about uh uh maybe let's say creative thinking. Industry is more on the implementation side. I'm not kind of uh going down on the

kind of uh going down on the implementation because that's important. But my purview of thought is or my my my uh intention of saying is that you if if somebody builds their foundations very strong then they can actually rise very quickly even though let's say somebody switched and got into a job let's say even with a higher pay than somebody with an Mtech did but what happens is that if you if the fundamentals of somebody is very strong then you know it kind of gets traction eventually like think Professor Andrew Ngji. Professor Andrew Njg invested uh his uh years in a doctoral program and today look at professor Andrew he's kind of into uh startups y combinators and uh education he has totally revolutionized

education he has totally revolutionized the education industry and not just Mr. Andrew but there are so many people for sure resolutionizing there are some good examples and sometimes we need to think that when we are talking about AI there are a lot of concepts that uh that like that are very difficult to master if somebody is just like kind of switching you know without any AI experience before. I see an Mtech program gives you the time to explore to learn concepts and maybe to build onto ideas because it's not just about cramming in the book but it also some kind of rigor. So as I said the stronger the base is in the in the long run in the base is in the in the long run an Mtech program can really help a student a lot to kind of if they

help a student a lot to kind of if they use it in the in the use it if they use it in the in the correct perspective. So I feel that both the options like you know switching as well as MTech they have their own benefits they have their own trade-offs. Somebody who is very much constrained on the financial aspect just wants to get on to the industry can make a switch because today we with the advent of online courses that we have learning is possible. But of course if we are thinking about very fundamental knowledge or some deep dive into subjects or you know kind of working onto some good thesis reports then I feel that at least in a domain like AI student without a prior background needs a master's degree because you see these days we have a bachelor's program being

days we have a bachelor's program being offered in artificial intelligence right I've seen that yeah and uh so very interesting which is very interesting because you know at the Btech level the student get so much of exposure and even I feel uh according to my subjective opinion a master's program even a two years masters is insufficient to kind of you know build on to so much of foundations because there is so much of a gradual learning but yeah I suppose that answers your question yeah I am curious though and I do want to this is intentionally to you know put you on the spot and force you to pick a side but what would you say to all of these people that say that oh you can do

these people that say that oh you can do all of this learning by yourself it's all available on YouTube you can you know build those really strong fundamentals the base that you were referring to by yourself if you're driven enough and you don't really need to enroll in a formal education degree for that is that not true then all right so let me divide your answer into three aspects so on the first front we have an a student who is let's say unemployed without any without any mast's program. On the second case, we have a student who is into a master's program. And the third case, we have a student who has switched and joined a company. Perfect. Now, let's think about their situations uh uh like try to analyze that because your question is a very interesting question and it's very

very interesting question and it's very obvious to ask this question. See, if I think about the third case, the student who has kind of switched into some other company. So what happens when you work in an industry you know most of the time you spend uh you know doing some implementation projects right so you have to debug you have to do so many things and then maybe you are in some another city so you have to commute you have to travel that's going to take up your good deal of time. Yep. Yep. So amount of time the the the amount of time that you are left with is not that much to kind of hop you on into deep

much to kind of hop you on into deep concepts. Right? Maybe you can be an AI enthusiast. Maybe you can do some professional program. But the thing is that if you really want a deep dive, if you really want to build strong foundations, then just if you're just only working in an industry and not taking up a serious academic program, I feel subjectively. So again, I'm not trying to assert onto anybody because I don't want to offend anybody.

don't want to offend anybody. No, no. Yeah. Firstly, I made you assert. So don't even worry about Okay. So yeah. So continuing with the answer. So the time that is left is going to be very less because see in a IML 100% you need to kind of be studying over the weekends cuz you you're so tired during the day. Exactly. And this just no time and yeah you'll you'll not want to go as deep and use your brain as much just because you we yeah we think about the mentality because the brain is spending 8 to 10 hours in the office and then lot of things happen in the office. Maybe for sure your colleague got promoted.

your colleague got promoted. So you feel like I have to do my office and then this becomes a secondary thing for you and in AI again if you if somebody wants to learn they need to kind of read they need to work out the mathematics it's not just kind of passive reading and it requires time investment. So that is one case. The next case is about somebody let's say who's not employed and who is even not in a master's program. Let's say somebody's kind of just studying. So I feel that for those people the target should be that they should try for a master's program but they should not think of it as this way that oh my time is getting lost. No if

that oh my time is getting lost. No if somebody has got some time they should invest in learning because if you do self-arning there are totally a lot of resources. I kind of write about about resources at LinkedIn. I get a lot of questions from students and I tell them that it's it's okay like you just need to uh study a foundational book you need to talk like I think the road map section we can discuss that later on but yeah so there is the process for it so there's nothing to feel bad about it the third aspect is the program now being into an Mtech program what it does is it gives you a lot more motivation because you are meant to pursue the program that is your primary mot uh priority So that kind of fosters learning much more than in the other two cases.

in the other two cases. Agreed. Y and basically that pursuing a program you know it gives you the foundations fundamentals. It gives you the backbone like you know it gives you a tag as well like some people like it to call it a tag as well. But uh and you're in the right rooms with the right people that are asking the right questions. Exactly. All engaged in research. You Yeah. It's just the perfect fertile soil for it's just environment that kind of makes the difference. Definitely. So, and you know as you mentioned I do obviously want to touch on the road map because it was one of the it is clearly so critical in this time and age as of recording this in in

time and age as of recording this in in 2025. Um but I do want to quickly take a like very quick detour and um you know just just I'm just curious Aayush when you were considering your you know next level up or so your M's program did you ever consider going abroad for your masters at all? Yeah this is a very good question and uh uh I did had I had to uh do some kind of self introspection. Mhm. Mhm. And uh it varies with the person to person. feel that uh uh both going abroad when just think from a master's perspective I think both going abroad and remaining in the country have their own benefits. Can you say your POV on why you chose to stay versus not go?

why you chose to stay versus not go? My yeah my uh reason fundamentally arose because uh firstly first reason was that at that instant of time I wanted to be closer to my family. That was uh one aspect. Yeah. Second aspect was that I was not really mentally prepared to make a kind of a investment on the monetary side for my master's program. Although I do not see that as a uh negative aspect uh but I but at that point of time yeah I just felt that yeah no it's definitely a factor for a lot of people cuz it's yeah it's like several orders of magnitude more more exactly and what I yeah exactly and what I felt is that pursuing a masters in India was like kind of a very good

India was like kind of a very good option in the sense that at a very econ economical rate given that I come from a middle-class family uh at a very economical rate and proximity to my house and uh I'll be able to build a strong launch pad. So that was what my main uh contention was because yeah I do agree with this fact that if somebody goes offshore the kind of opportunities and the outreach that you get is more than what you can have in your country. But uh I feel that if with rigorous networking and uh kind of uh learning spirit and zest maybe we can match or perhaps even outpace somebody who has gone offshore. But I think again for me uh these were the two factors

for me uh these were the two factors like proximity to family and a financial consideration and uh that was the main reason because I was thinking from a launchpad perspective. I wanted to see that if I invest in this uh in this domain where would where can it possibly take me. Yeah. So considering those aspects, I felt that okay, India is going to be a good option for me because I did give my GRE, I did give my TOEFL, I had my scores as well. But at that point of time, I just felt that okay, let me go ahead in my country and explore. And also there was this IIT craze a bit as well, you know.

this IIT craze a bit as well, you know. Yeah. like you always feel like you Yeah. like you always feel like because during my J days like I did well in webg I got into JU but still people kind of say that oh you didn't got into IIT so I kind of somehow felt that okay like coming into a mast's program in an IIT maybe kind of completed that effort that that I efforts that I made but again I feel that my decision was there and it was like not very hard bound like it was not very much objective.

it was not very much objective. Uh I just felt that okay let's I mean most decisions that most people take are not objective. So objective. Yeah. Yeah. I just feel like that's totally normal. I I am curious though given how well-versed you are with you know curriculums in the west and uh you know as you said teachers in the west and how stuff goes around here. Um would you say that I guess yeah what has if you had to candidly describe your experience while at uh IIT Karakpur getting your MTech how has it been? Has it was it everything that you thought it would be? Were you disappointed liked it? Just looking for your um experience essentially. essentially. Okay. So yeah this is a bit difficult for me to answer. Uh-huh. And uh uh to

for me to answer. Uh-huh. And uh uh to be very frank, I would say that it has been a good experience at IIT, but it has been an experience of a lot of hustle. I would say because about that. Yeah. So when I joined IIT Karakpur, so I came from a electrical engineering background and I did had some a IML fundamentals prior to joining because I had given the gate DA paper as well. So in IIT Karatur like when uh I was in my first year first semester we had course work on machine learning linear algebra and then but the thing in IT Karapur was they were offering depth subjects in the first semester itself. So generally what happens is that you kind of only take

happens is that you kind of only take breath co like the foundational courses but the way the program was designed so what happened was that we hustled a lot but we ended up learning as well uh in proportion to how much we hustled. So there was it wasn't hands-on. Yeah. Was it like did you have to build projects or was it kind of like of course of course there was a lot of theory to it because I kakpur is very rigorous in theory when it comes to building theoretical foundations and of course in every course we had like kind of one or two course projects like I still remember in visual computing class I we built a gable filter and people did very fancy things and uh in every course

very fancy things and uh in every course I mean even in machine learning and the weight was done like for every coursework we had a project and everybody used to be there and present it was just like a thesis presentation for a course work and you know the profs used to say that everybody should attend and listen so it was a very interesting experience it was a very illuminating experience for me but again the course work was sometimes very overwhelming for me because I think uh how much you learn is decided more by what subjects also you take uh there's an option to evade a breath course with let's say some different course let's say some somebody takes a happiness course or some

takes a happiness course or some different courses because they want to get a good CG that also you can do but then it's up to you like what coursework you want to take because the department is ready to offer you all types of courses from all breads from vision to NLP to graph and it's up to the learner and what I feel is that yes apart from the coursework what an individual needs is a lot of self-study because AI is a domain where just by attending ing the classes it's not going to help much. One needs to invest a lot of time doing self-study, understanding the mathematics, working out the mathematics, writing the code. You know, you read a research paper, then you try

you read a research paper, then you try to replicate it and a lot of times you get a lot of errors and then you have to backtrack. Maybe you're a reverse engineer and you have to think about a lot of aspects as well. So in my uh opinion I feel that IIIT Karapat has like very strong professors in AI because I we had a professor on visual computing she did her PhD from UT Texas Austin uh under a very legendary prof and I think most of the faculties they are highly qualified and uh if a student really kind of takes the initiative a student can really learn a lot in this program but of course it's going to be a bit hectic of course uh because given the fact that in most other IITs, IIT

the fact that in most other IITs, IIT Kadakpur like we generally take one subject extra because as I got to know from my peers like in other IITs in a sim they are able to take four courses but in I think KGP we have five theoretical courses we have I think one or two lab works and we have a seminar as well so on that regard of course a student can learn a lot but yeah there will be a lot of moments when you you'll hustle when you'll think that did I do it correct by taking up a difficult course. So, but of course in my opinion, yeah, the learning options are good and uh I have not taken other programs from other IITs, but of course uh given my experience at IIT KGP, I feel that if a

experience at IIT KGP, I feel that if a student is serious and willing to learn, they can definitely learn a lot in this program. program. That's super helpful. Yeah. And obviously you know in the short time that I've known you, you do come across as somebody that takes learning very seriously and I know that you would not be saying this if what your intention was uh before joining if that had not been fulfilled right now you would be saying a very different answer right so that that that makes me really happy to be very honest with you Aayush because I feel like and I don't have a master's degree from India right I obviously came to the US to get mine but I find that and this is just I feel like just unfair in a lot of ways but people still feel

in a lot of ways but people still feel like the like Indian education system does bleed into the master's courses as well like so at least you know I maybe JU does it differently but for my bachelors in engineering there were no there were next to no projects there was no like hands-on learning you just had to study the week of your exams and you would be fine but it really sounds like in your experience it has been very different from that that is no longer the way and that though happy to to hear from you.

though happy to to hear from you. Of course. Yeah. truly because uh at Mtech program at IIT Karakpur definitely somebody thinks about because we have projects and we have assignments and assignments not just really assignments in the name of formality but assignments that really make you think the course because what I believe is the most important thing to learning is consistency and consistency is established by the way the coursework is built and the courses that I have taken in IIT Kadakur the way they have been built they kind enforce this consistency aspect. If a student is not consistent, of course, their grades are going to take a setback on that and of course the learning as well. So yeah, I think that answers that. Yeah. So, and if I had to press you, uh

Yeah. So, and if I had to press you, uh what is maybe one thing that you according to you could have been better uh during your time at IIT Karakur? Yeah. So, it's it's it's a very nice question because it uh uh leaves more room for improvement. Yeah. I don't something right there's all all of course definitely I feel that uh basically uh uh of course the it's like we have some really nice state-of-the-art courses but one aspect that I would really touch upon is that uh you know I mean the way the courses are arranged basically let's say I said so you that we to we had the depth option in the first s maybe if the depth option is happens to be the next s and

option is happens to be the next s and more courses on the foundational part in the first step then maybe got it. It becomes more easier to manage the courses. courses. Oh yeah. because that was one thing and uh of course like uh introducing uh we can have like more courses like for instance let's say that convex optimization is not considered a core course for MEC but I think it should be a core course for MTech because in BTE for AI it's a it's a core course and uh basically yeah like it it the program the program can incorporate more mathematical courses as well but there is like ample amount out of mathematics but uh like it can make it even more flexible like room for makes sense allowing students to either delve into

allowing students to either delve into more mathematical side or into more application oriented side but I feel that overall it's a very good course because of course the profs who uh the profs who like frame the department like professory founded the department and he's a very renowned professor in NLP and social networking so I feel that yeah it's a very good program and uh again like whether a program improves or not is not just guided by the program but also by the student as well 100% 100% like% the way the student wants to cater his learning if a student is like takes up a course and then maybe later on revises it and tries to apply it so he will have been a better experience than somebody who just take courses for

somebody who just take courses for granted and just thinks that okay I'll just in the end the placements will come just have to to get something. So I think it also boils down to the mindset of a candidate as well. So I think that yeah no that's such a great call out and it does make such a big difference. Yeah I really appreciate you being you know so candid and as candid as you have been jumping onto the road map you know here probably most people are the most interested in this particular section.

interested in this particular section. Um so if there is somebody in their third or fourth year of engineering they want to build a strong base in MLAI uh you know theory or maybe catered towards implementation but at least I know that you have to start with the theory. Um what are some of the the best go-to resources out there that you that you like to recommend for these people?

like to recommend for these people? Yeah. So let's uh uh try to analyze this question um in a bit more detail. So I'm trying to look upon the engineering branches of the student because that really matters a lot. Let's say we have a student from a computer science department uh who is in his third year of engineering and let's say we have a student in ocean engineering or let's say naval architecture or let's say civil engineering. They are very fascinated about AI. So how do they also kind of approach it?

how do they also kind of approach it? Yeah, I'm so glad you included that. Yeah, that's such a great addition. Yeah, Yeah, because there are a lot of other departments as well and given the fact that AI is so ubiquitous, everybody wants to hop into it. So let's think this way. Firstly, a student who is in their third or fourth year needs to be very clear what their goals are. First is that do they really want to build very strong foundations or they just want to get a job after their B tech. Got it. So if they want to take very strong foundations then again the question diversifies is are they from tier 2, tier three or tier 1. Yeah. Yeah. So if you are from tier 2, tier three definitely you will write GATE DA. So

definitely you will write GATE DA. So you need to prepare for GATE data science and AI paper. If you're from tier one, you might also consider preparing from GATE DA if you are from a nonCS or AI branch or you might think about going abroad. But let's now come to the materials first because they need to prepare for the competitive exams. First thing of course is the mathematical foundations that that we need to talk about. So machine learning.

need to talk about. So machine learning. So machine learning is basically when we had this AI thing. So AI actually had three divisions like one was logic, second was computation and third was probability probability and in probability you had the probabilistic machine learning. So the entire domain of machine learning is stochastic. You know it's uh it's a stoastic approach. So the first uh thing that students should be very good at is understanding probability and to learn probability they have to like consult uh for I I always recommend students to read foundational textbooks because that builds the thinking ability but of course there are some good lectures on the internet as well.

lectures on the internet as well. Yeah. So for instance when we talk about probability there are some good materials by this uh MIT open course where professor John Cyclist has very good lecture on probability theory then there's a very nice book on probability by Sheldon Ross okay and uh so those are the things that students start to prepare with and they don't need to read the whole book basically because of the time constraint what they need to do is look at the gate syllabus and see what are the important topics because it's not only just about exam preparation ation but it but the topics actually include those things that are important right so if you study those topics you're not just only catering for the exam but you're also preparing for the interviews because the

preparing for the interviews because the kind of questions are anyways going to be very similar so first is the probability fundamentals you need to build second is that they need to learn linear algebra very well because in machine learning you have lot of tensors you have a lot of matrices operation you know you have normal equations in in this linear regression you have principal component analysis you I get values I get vectors. So for linear algebra there's a very good lecture uh material by professor Gilbert's trying.

material by professor Gilbert's trying. I think a lot of students know about professor Gilbert's trying a very famous professor at MIT. He gives very nice intuitions about building linear algebra. And then a third part give all of these in the description for anybody trying to find all of this. It'll be those show notes. Yeah. But sorry let's continue here. Yeah, because this is something that I've written uh in London as well the materials because lot of students ask me this question only and then you have the calculus because you have uh you you have back propagation you need you need to be very well with Jacobians Hessians and uh calculus as well. So you have the gradient calculus. So these three are the very important math foundations probability junior algebra and calculus and if a student is wellversed in these

and if a student is wellversed in these things then they will be able to understand because mathematics is just like the language of machine learning. Machine learning engineers speak via mathematics and if somebody has a good grasp over mathematics they can understand well it's just like we we speak in English or Hindi and the better somebody has a good grasp of it the faster they can understand. So math is just so math just dictates how well can you understand the AI or machine learning. So that is the math aspect.

learning. So that is the math aspect. The second aspect that we come to is the non-math aspect and here is where most of the uh non CS folks get a bit scared because you know they need to prepare data uh data structures and algorithms foundations. foundations. Correct. And uh because this is something that is required coding is something that is pervasive and uh we need implementation for our models for our algorithms and we need to write code when you write code in PyTorch and other uh frameworks. So for that basically students generally what is expected out of a student in a gate data science and AI paper is to have good grasp over data structures and algorithms and for that

structures and algorithms and for that again there are two very well-renowned books. There's one book by Corman uh CLRS I think it's a very popular book and there's another book by Kleinberg and Tardos that was very good fundamentals in uh in in in in data and algorithms and of course from the job perspective what I have seen is that many companies they kind of consider lead coding as a crown yeah basically competitive coding so you need to like be also have a good grasp over lead code like maybe you should First study the a bit of books and then maybe solve the important design patterns in the lead code like you have the uh the the section explore section where you can we have explore cards or maybe you can take a crash course or maybe blinder 75 or

a crash course or maybe blinder 75 or grind 75 I mean they have a lot of names these days but of course yeah the students should know the pattern because in any ML interview that is the first route so there's always going to be a DSA question whether it whether you just do hardcore ML or whatever but uh data set and algorithms is like asked because they kind of correlate uh companies correlate uh that with the aptitude of a candidate. So they have to prepare for the DSA as well. And of course there are other concepts like database management systems, DBMS, SQL those things. I think there are of course standard materials and students who are from non CS background need to invest some amount of time in learning because that might not feel very pertinent to like ML theory but that

pertinent to like ML theory but that becomes a important component when students are going out for interviews and placements and those aspects. So I think that answers that because together the math and the code and uh this is what I would say for a B tech third or fourth year guy. Now if they're very much interested more interested they can dive into research papers as well.

dive into research papers as well. Oh wow. And uh in research now research papers again they have a lot of research papers like so what's important is to first of all look at the topic that you are interested with and then maybe look at the some of the very renowned profs who are working on that and then maybe look at their papers their uh seinal papers maybe like for instance in machine learning what were the some of the seinal papers like I know transformers is all you need was one seminal paper Alex that resnet was a they were very seminal papers So they can try studying the papers and uh you know kind of replicating the those papers. And another important aspect is that having a foundational textbook like students should have a foundational textbook like it could be uh the book by

textbook like it could be uh the book by Christopher Bishop or it could be the book by Goodfellow or it could be like I say that it can be uh any book but it should be a good foundational book that should students should stick to it. So I think these are the aspects that together culminate amalgamate a strong preparation and of course students to invest in a lot more time to build uh to build their fundamentals because AI is something in my personal experience is it it's a subject that demands a lot of time from you and uh it's just like the value of despair kind of thing where students put in a lot more effort and

students put in a lot more effort and say that man I'm not understanding anything and what's going on So those moments will always be there in in in the learning phase and uh the hard fact is that any learner has to overcome that. So I think yeah that kind of answers your question for sure. Yeah. And to to the your time point, I just want to add that as with most things in life, right? The graph is just like this is the good and bad effect of compounding where nothing happens, right? It's just a flat graph and then suddenly you see, you know, like a huge spike up after a really long time, but that only happens because of all of the time that you've spent in the

all of the time that you've spent in the past apparently not making any progress. So, I do like that you called that part out. And then secondly, I am also curious if you had to put a range on it, roughly how many hours would you say somebody can have in their minds before at least they can start, you know, cracking the first round or and such. Um at for these jobs, if you can even is is that a dumb question like can you even is that? No, I think I I think yeah, I think this question is something that estimating the number of hours is very difficult because again No, I get that your Your question is good in the aspect that you're asking from a round perspective but yeah I

from a round perspective but yeah I think the number of hours that somebody needs to invest varies a lot like I feel that it depends on the focus level of a student it uh depends on the environment it depends on how proactive somebody is so I think it's consistent they are also right how consistent they are because you know if somebody studies let's say at b 2 3 days and then they leave it without studying anything then it doesn't make that much of a progress. Yeah. Yeah. But even if you're investing 2 hours or 1 hours of your time learning and you do it for consistently for an year, then I think that compounds it. I mean is it's just like 1.01 to the power of 365 is

just like 1.01 to the power of 365 is like 37.7. Yeah. And so it's just like that kind of uh that's Yeah. I think that Yeah. I feel like that actually answers it even better than any number that you would have given just because it doesn't matter, right? like go do your work. Make sure you put in your work, be proactive and uh have an enthusiastic mindset and that's that's more Yeah. Love it man. See this is my thought perspective you know it's not that I've looked it somewhere. somewhere. No no but yeah and but this is but because you are so deeply embedded in this space I would any day rather ask you than you know was some of my random friends that have nothing to do with this. So it you're like I really appreciate you calling out that this is

appreciate you calling out that this is not like don't quote me on that essentially but it's super helpful for me and my listeners because yeah a lot of us are not as deeply embedded into tech as we would like to be. So it is always such a pleasure and privilege to hear from somebody that you know not only takes this stuff seriously but as you said you know you're more focused on the research the knowledge part of things more so than you know the implementation necessarily which I sue lair in today's world I feel I feel that uh we should not see any kind of application or anything as a black box we should try to dig into it and see how it really happens and I Yeah,

it really happens and I Yeah, that was the major shift in my thought that kind of propelled me to an AI program. Otherwise, why would I have kind of hopped into it? I could have just made a switch or you were in the interview already. Yeah. So that is the main thing that I think it depends on the perspective like maybe somebody might not want to go into the nitty-gritty of it but I am the kind of thing like having studed engineering and you know being curious curiosity drives me a lot like I get curious uh if there's something new coming out uh like in LinkedIn also so many things like the thinking machines thing that uh thinking machines lab you know uh some months back they claimed that we can get

months back they claimed that we can get deterministic outputs now consistent outputs because of the the even the temperature parameter doesn't makes the LLM get rid of stoasticity but they came up with some way so things like that I mean you know it's very important to to kind of break down that black box and see how things are working because if we really want to do well in something we need to appreciate it that's very important important it's just like reinforcement learning reinforcement learning says what that you do not optimize the current reward maybe the Current reward gives you very high reward but you optimize for the value function and the value function gives you the expected return over a long period of time. Wow. So so that's how that's how because RL was

so that's how that's how because RL was kind of based out of human idea. So as maybe as humans if we are getting very much shortsighted if we are looking at instant gratification maybe it's now time to look at AI maybe look at RL principles and implement that back in our life. maybe complete the feedback loop. loop. So maybe we should think no that that that is so incredible. Yeah, that also worked as in dopamine and things like that as well in in that same analogy which is beautiful. I think I think it it is I mean it is tempting to try out new technologies but of course building foundational principles are very important. Oh this has been so incredible. Truly this has been per minute one of the most

this has been per minute one of the most information dense podcasts I have had the pleasure to record. So truly thank you so so much for taking the time not just today but also on LinkedIn for guiding so many of our country's youth towards bright and helpful resources and truly I I I'm walking away with so much new so many new insights and I'm sure anybody listening would have also. So thank you so so much Aayush for taking the time here today.

the time here today. Thank you so much Nan. That brings us to the end of that fascinating episode with Aush. If you found value and if you would like to support me, the easiest way to do that is by subscribing on YouTube and leaving me up to a fivestar rating on Spotify or any of your favorite podcast apps. Something that goes a really long way for me is if you share this episode with somebody who might benefit from it or with your friends and family. Catch you all in the next one. New episodes every week.