Hit Refresh : A story of purposeful resets | Yogesh Kulkarni | TEDxCCOEW
URL: https://www.youtube.com/watch?v=-VbWRs7BsPY Video ID: -VbWRs7BsPY ============================================================ Transcriber: Dương Chu Reviewer: Kiều Anh Nguyễn Many of you may have heard this phrase “Change is the only constant”, right? I’ll go further and say change is not just constant, but it’s accelerating, especially in these days of technology, and that to artificial intelligence. Things are going to come very rapidly to us. In my previous generation, you would start with a job and continue it til the retirement. In my generation, you would change jobs after a few years here and there, Over the career of 20 years, you would have changed 5-6 jobs, right? But still, job changing is not really considered good. It sure will be frowned upon and especially if you have gaps in the middle, right? But these things are going to change. AI, technology, and other external factors are going to force upon these changes on your career. So I’m going to talk about the resets, the changes that you may have to do in your professional career because of these technological advancements. Why am I saying this? Because although these changes are going to come in the future, I lead them in my career itself. So I’ll be taking some examples from my own career to explain this point. Let me give you my background. I’m a mechanical engineer and in mechanical engineering, we have a subject called machine design. What is machine design? Design of machines Basically you design, say automobile, aeroplane, parts and simple things like a table and chair. How do you design a chair? You basically decide its dimension, material, thickness and other things, so that if somebody of, say, 100kg, 120kg sits on it, it doesn’t break. That’s the basic idea. But with the advent of computers, you don’t have to manufacture many chairs and then test them on a prototype. You do it on computers, you do it in software. And that thing is called as CAD - Computer-Aided Design. And you test it in something called CAE - Computer-Aided Engineering. And that was my favorite subject in mechanical engineering. So I did Bachelor’s here in Pune, and then went to the US for Master’s, and I did Master’s in CAD. Worked there for a year, came back here for good and then worked again in the CAD domain itself. So I was working in a big multinational company, MNC, which had a parent company in US. Typically that's the setup, right? You would have a European or American parent company and an office here. So things were going fine. I started with the software engineer, senior software engineer lead, working in CAD domain. But one day, I came across one ad where somebody was going to start a startup. Somebody from the US was planning to come here and going to start a company. Few peculiar things about that startup. Like typical, you have offices in Europe or America, this company had office in China also. So that was a special thing. Another thing it had startup nation. Nobody was there. No employee was there. And then the guy was coming here to start the whole thing. Another big thing was, it was working for a big giant company, number one. So looking at all these things, I decided to take a plunge. So that was my first reset. First reset in the career. So I decided to leave my cushy job, and started to join, or decided to join the startup. So started with that startup. First day. I was the first employee, one more person joined, and we started in somebody else's office. We had two tables at the back. No HR, No IT. Previous day I had all the facilities. I had people to assemble my computers. They were HR people. But here nothing, we were on our own. We started the journey like from scratch, like a reset. We did worked well, I think. Eventually that CADer giant acquired us. So we became the CADer We worked on projects. We did reasonably good work. And then I went from, say lead to manager, group manager, site leader. So journey was very fine. Professionally, things were looking good. I was in a CAD giant company, possibly the highest position possible in Pune in that domain. But something was missing, and that missing part was on the academic side. So I had Bachelor’s and Master’s, but not PhD - the highest possible degree. So I talked with my American manager, said that I want to do PhD. So I enrolled in College of Engineering Pune for PhD. In PhD, typically you will have to do coursework initially, and also do what is called as gap analysis or literature survey, that I did while being in job. So I held all the responsibilities of the group and also do research. My aim was to do very meaningful PhD, not like any X, Y, Z PhD. So I thought doing both things were not possible. The responsibilities were there and doing meaningful research was not possible. So talked with my manager and decided that I should take my second plunge. I decided to leave the job and become a student again. So previously I had position, authority, money. Everything was fine. Next day I was a student. No money, no authority. And I was studying with the guys almost half my age. This was the next reset that happened. So actually the PhD is in, again, CAD domain. So decided to publish a paper every six months. And that was going fine. So I was working on a module called “Defeaturing”. Let me explain what that is, very simple example. If you’re designing a table and then you want to test it, both things are happening on software. On a table, if you want to do testing, again, you have to simplify the shape. What do I mean by simplification? If there are small holes, fillets, chamfers, you typically remove them, simplify the shape, take it to CAE for machines and other things, and there you test it. So this removing some small things called features, is called “Defeature”. That’s the module I was working on, like a part of research. Writing rules “If the radius is less than five millimeters, remove it”, “If it is less than ten millimeters, remove it”, so on and so forth. And this research is going on for many, many decades. You need engineering judgment for it. But I came across one paper which took a radically different approach, altogether a different approach. And that approach was, instead of writing these rules of software logic, that guy gave original shape and defeatured simplified shapes to some approaches, (with) hundreds of such examples. And that approach so-called miraculously found out the logic of “Defeature”. This was an “Aha!” moment for me. This is miraculous. And this is what is known as machine learning, supervised machine learning. It is a part of AI - Artificial Intelligence. I was enamored by the whole thing, so I decided to take my third plunge. With 20 plus years of experience in CAD with a PhD in CAD, I decided to leave CAD and started with machine learning. As I finished my PhD, I became a machine learning researcher. At 40+ age, nobody will give me a job with that seniority with that salary package. Not affordable. So I started working for some small startups just to build my portfolio, without money, without taking any money. I worked there for a couple of projects like that, and then started getting paid projects also. Eventually got a job in that domain, worked there, became principal architect. But then sort of thought that once you are in a company you are sort of bound by what that company does. So if the domain is something, you only get AI in that domain itself, right? You don't get anything else. I decided to take my fourth plunge. Decided that even if I have this very cushy nice job, I should become solo again. So I left my job and became an AI advisor. In AI advisor, you have to scout for work. Eventually I got good projects, different domains, finance and all, and then working in that domain. Things are going fine and I’m now waiting for another reset that will come probably in the future With this as a role, I would like to explain and suggest new things to new generation. So people consider AI for this common sense guys, mathematics guys. That’s not the case. It’s for mechanical engineers, also. It’s for electrical engineers. Civil engineers also. Not just engineers. It's for legal guys, medical guys, even guitarists. Everybody should be aware of AI because it’s not up to you. It is becoming mandatory, compulsory to know at least something about AI. I would categorize the way you use AI in three roles. So I’m sort of now suggesting two things to the new generation. You have to use AI. That’s given. But in which roles, right? You can either be a user; simple user. How do you get prompt engineering everybody is aware of, how to use it, that kind of thing you can do. Second thing is if you are a little programmatic, little developer-oriented, programming, then you can be a developer. You can build apps in AI. And if you are really good at mathematics, then you become researchers. You invent things in the end. But you have to take one of these three roles, whatever your field may be. And the idea is, I would suggest not go for only AI. That’s my career. That’s not the case. AI is actually a tool. You have to have your own domain. If you are a mechanical engineer, be a mechanical engineer. Solve your own problems. If you are an artist, have your own problems. But use AI, all the time to solve the problems As the AI advances, newer and newer models come every day. Apply them. Do they work for my problem? Well? Then good So what is happening because of this combo? You become specialized, right? You are already aware of what’s going on so that advantage is there. Plus, you have not left your domain, so that’s the expertise nobody can steal from you. So having this combo, it is going to work wonderfully. There is something called Gartner Hype Cycle. Some of you may have heard it. The technology grows. Everybody starts using it. It’s at the peak. How do you decide it’s at the peak? Everybody starts talking about it. That’s the peak. That’s what is happening for AI. But what happens? Everybody starts using it. Some things work, some things don’t work. Then the popularity goes down. So it's a cycle like this. It goes down and then wherever it applies, wherever it works, that maturity comes and then it becomes stabilized. So currently, AI hype is going on. So without leaving your domain, you have to try AI-related technologies in your domain, see if it’s useful; if it’s not useful, forget it. Newer and newer technologies will come in the future. Your domain has to be there with you, as your expertise. So my suggestion is be on the AI wave. The theme of this TEDx is Tarang - Wave, be a rider on the wave - Tarang Savar, or Tarang Mitr - be a friend of the wave, not oppose it because you don’t have a choice. So looking at the examples from my career, I decided that I should title my talk as “Hit Refresh”. You've heard of this book title, right? Very famous book title. Hit “Refresh” to refresh your career, at least for the newer generation. Externally or by design. I did it by design purposefully, but you may have to do it by external factors. Hit “Refresh” and start from scratch. Your career path cannot be like this, like what we had in the previous generation. It cannot be sinusoidal, smooth, like this. It cannot be. For me, it was sawtooth wave. Resets all the time. Just that you have to see. The sawtooth wave is going up. Right. And it’s not going down. That’s the idea you have to do. Take care of it. So whenever in your career, if you see good opportunities for change. The change should not be incremental. Plus 10th percentile, 5th percentile. That’s not the change. You have to have dramatic change, orbital change. You have to go to the next orbit. Then will you change. That's my suggestion to all of you. As I borrowed the title of the talk from another very famous book title, I would summarize my talk with another borrowed line from another tag line. So whenever you see change in your professional career, smell a change which is of orbit change. Then I would suggest just do it. Just do it. Thank you.