Please tell us about yourself.

I’m an Artificial Intelligence (AI) Engineer at Kindred Systems Inc. I’m a part of their AI Research Team in Toronto, Canada. We’re working towards building machines with human-like intelligence. Our goal is to develop fundamental principles of intelligence and designs of general-purpose artificial minds. My research interests include Deep Learning, Reinforcement Learning, Robotics, brain and behaviour.

I graduated from NIT Trichy with a B.Tech in Instrumentation and Control Engineering in 2015. At NITT, I worked on control strategies for a robotic vacuum cleaner for solar panels and autonomous navigation for quadcopters. I then went on to pursue my M.Sc in Computing Science at the University of Alberta, Canada. I successfully defended my thesis in August 2017 after working on reinforcement learning methods that would allow an amputee to use their non-amputated arm to teach their prosthetic arm how to move through a wide range of coordinated motions and grasp patterns. This was a fully-funded M.Sc program and I won the 2017 M.Sc Outstanding Thesis Award.

How did you end up in such an offbeat, unconventional and cool career?

Robotics started as a hobby for me when a couple of friends and I decided to explore in our first year of engineering. It later blossomed into a passion which led me to a fully funded M.Sc in Computing Science at the University of Alberta. I graduated from NIT Trichy with a B.Tech. in Instrumentation Control Engineering, where I was lucky to be involved in multidisciplinary research. The decision to defer employment and join a master’s programme abroad was easier for me since I had had a taste of excitement and the sense of fulfilment obtained from research. Considering the huge financial investment required, I applied exclusively to top-tier universities in the USA, Canada and Sweden which offered funding/scholarships to their graduate students.

Original Link:

http://www.thehindu.com/features/education/Passion-for-robotics/article14580466.ece

Why Alberta?

The University of Alberta is consistently ranked among the top four universities in Canada. It has a stellar reputation in research and development and has world-class research laboratories. Initially, I struggled with the coursework as its structure demanded delving deeper into concepts and encourages one undertaking projects. This was drastically different for me since we were trained in the “art of taking exams” rather than building our own project ideas in India.

How was the experience?

The great thing about the university is that all graduate students involved in thesis-based programmes get paid to study. The stipend covers both tuition and living expenses. Earlier, I was a Teaching Assistant for an undergraduate course on programming which helped me learn about their teaching methods, student perspectives and so on. I’m currently a Research Assistant in my lab at the department of rehabilitation medicine. My research focus is on reinforcement learning and robotics — creating intelligent artificial limbs to restore and extend abilities for people with amputations.

Weekly seminars by graduate students, invited professors and key people from the industry give us a glimpse of the exciting research going on around us. In addition, there is excellent funding for research and many opportunities to attend conferences and present results. Professors are extremely supportive and always willing to listen to your ideas.

Life in Edmonton?

The friendliness of Canadians, and especially the people from Edmonton, has left a lasting impression on me. Edmonton is a beautiful, ethnically diverse city with low crime rates and taxes. The winter is a bit harsh as the temperature sometimes drops below minus 20 degrees. But I started to love the snow as soon as I bought the right winter gear. Summer and spring time is so breathtakingly beautiful along the river valley that people try to spend most of the day outside. It’s easy to obtain Indian spices and food thanks to the huge Indian diaspora living here. Hiking, camping, canoeing, skiing, snowboarding, etc., are some of the enthralling activities that are common here.

My life in Edmonton has given me a sense of direction, confidence and freedom. It’s helped me prepare for an exhilarating journey ahead!

What should students know before taking up research? How should they pursue their career in research?

Intellectual curiosity and perseverance are extremely important. Prepare to pour your heart and soul into an experiment only to see it fail in spectacular fashion. The key is to learn from your mistakes and improving yourself every single day. You are the average of the five people you spend the most time with. Therefore, choose your friends and colleagues wisely. I can honestly say that I’ve never been the smartest person in most rooms. That’s quite a liberating experience. It forces you to stay outside your comfort zone which in turn helps you fail fast and learn faster. Over the years, I’ve started to be more cognizant of my own inadequacies and gaps in knowledge. Being a researcher can be a uniquely humbling experience.

Science is a cooperative venture. It’s about sharing ideas and building upon the treasure trove of human knowledge collected painstakingly over centuries.

One of the most underrated yet most valuable skills is the art of communication. Learn to clearly articulate ideas in a clear, concise fashion. Strive for simplicity and elegance.

Growth is incremental. If you like a topic/research area, try working on some small projects in that area. If it’s still interesting, work on bigger projects and try working with professors who are active in the field. If you’re unclear about your likes and dislikes, don’t despair! College is the perfect time for exploration. Checkout interesting courses on edX, coursera, MIT open courseware, Udacity, etc., and keep an open mind towards learning.

Why are research internships important? How do they help students hone their skills? Can you describe your experience in getting an internship?

Research internships are the perfect time to explore different areas in your field of interest. It exposes you to state-of-the-art research and different ways of thinking. I started reading technical papers during my first internship. It takes a while to get used to the technical terms used in the field and understand what’s actually going on. Reading a lot of technical papers is an integral part of research. We learn better under the guidance of good teachers/mentors. It can be overwhelming to start from scratch in a new field. Great professors ask the right questions, exhibit expertise, and spend time continuing to gain new knowledge in their field. They instill a hunger in their students to learn more on their own.

Getting your first internship is quite challenging. I remember mailing atleast a 150 professors all over the world in my quest for a research internship (2nd and 3rd year combined). All my emails had my resume, a cover letter briefly describing my research interests, my achievements and what drew me towards the particular professor’s research (I usually talk about their recent papers). You should spend a lot of time going through university websites and analyze the work of people you’re interested in working with. For international internships, given visa restrictions and processing times, you must apply atleast 6 months before the start date. If I remember correctly, you must apply approximately 8-9 months before the start date for DaaD and Mitacs scholarship programmes.

What other skills do you think students need to learn before/while pursuing higher education in engineering and sciences?

This is hard to answer. There’s no blanket solution that works for everyone. Since my 2nd year in NITT, I spent all odd semesters working for Festember and even semesters working on research projects (with some involvement in Pragyan and NITTFEST). While NITT left me woefully unprepared in the academic front, my stints in Festember (2 years in the Marketing team, Treasurer in my final year) greatly improved my interpersonal and public speaking skills. It instilled a sense of confidence in my own abilities. It helped me become a more rounded individual. Thanks to Festember, I usually did a fantastic job in my non-technical interviews. As a Research Assistant at the RLAI and BLINC Labs in the University of Alberta, I had to demo my work to multiple dignitaries ranging from the Mayor of Edmonton, Canadian Ministers to Professors and high school students. My Festember experiences were instrumental in honing my presentation skills and elevator pitches.

From an academic standpoint, learn to be self-motivated and a good team player. Enthusiasm and motivation are highly contagious. It helps a great deal when you undertake challenging projects.

Where do you think, the research scenario in NIT Trichy needs improvement, and how can it be improved?

The current research scenario in NIT Trichy is abysmal. We can talk about improvement once our research standards are atleast on par with older IITs and IISc Bangalore. How can good research occur in an institute which places higher premium on attendance, dress codes and patriarchal, outdated moral values? The entire system is geared towards churning out clerks rather than foster intellectual curiosity. A huge amount of time was wasted on writing observation and lab records (2 written copies of each experiment performed with equipment that doesn’t work half the time). An entire class would submit the exact same copy for assignments and no one bats an eye. I did have a few positive research experiences in NITT when I worked with Dr. G. Saravana Ilango and  Dr. V. Sankaranarayanan affiliated with the EEE dept. But I also had to suffer the wrath of vindictive professors who’d rather go out of their way to stymie good research than support us in our endeavours.

This is not an easy problem to solve. It’d take years if not decades to witness some progress. Our professors are incentivized for good research, but never punished for complacency and gross incompetence. Most are perfectly content (if not happy) with mediocrity. Respect is earned, not given. It’s fairly obvious that the most respected faculty in NITT are also the most intelligent, hard-working and humble. We need professors who publish regularly at top-tier international conferences and willing to equip the students with the technical skills required to succeed in research.

To current students – You have to fight every single day just to keep your hopes and dreams alive. It’s no use ranting all day if you’re not doing anything about your problems. Don’t get discouraged by faculty, work with your friends on interesting side projects. We have some of the brightest minds in the country ready to learn and take on challenging issues. You can achieve big results just by working on hard problems you are passionate about. If you have a penchant for research, I’d advise you to go against the grain and not focus on campus placements. Instead of taking a job that could derail your career prospects, take up internships and research assistantships with good research groups in India or abroad.

What must one do after getting admitted into college? I.e How do you feel time must be utilised post-admission?

This might be controversial. But you made it despite the odds stacked against you at NITT. Go celebrate! Drink, dine, do whatever you have to do to enjoy that moment. You deserve it! Once you’re past the euphoria, get back to reality, compare offers and make an informed judgement. This could be a huge, life-altering decision. Get in touch with alumni from NITT, current students at your post-grad university to get an idea of the school, quality of research, job scenario, etc. If you’re headed straight for post-grad from NITT, thoroughly enjoy the summer. Apart from paperwork and visa applications, you’d have a lot of free time in hand. If you are working, I’d recommend quitting at least 2 months in advance and spend your vacation travelling across India.

Keep in mind that not everything is star-spangled awesome in America. Each country has its own set of challenges and issues. You’d just end up trading your Indian problems for Western ones. Read a lot of blogs, learn about their culture and try to be a part of it. Remember, when in Rome, be a Roman. It’s no use moving abroad if you intend to hang out exclusively with the Indian community.

What are some things about your career path you wish you knew in college, in retrospect?

It’s okay to not know what you want in life. It’s better to take time off to figure things out in life rather than jump head first into a job you dislike from the bottom of your heart. While healthy competition is good for you, do not over analyze and compare every single aspect of your life with someone else. If you’re interested in a topic, start working on it immediately. Once you postpone, it’s unlikely that you’d ever work on that topic again.

Hindsight’s usually 20/20. While I was interested in AI, all my research opportunities were in classical control for robots. I was scared when I applied to multiple masters programmes in Computing Science and Robotics. Luckily for me, University of Alberta trusted my research aptitude and took me in. Had I worked with researchers like Dr. Balaraman Ravindran at IIT Madras, I’d have had a huge head start in my Reinforcement Learning education. Also GPA matters a lot when it comes to post-grad admissions!

What exams do you think are most important to write before applying for higher education?

GRE and TOEFL if you want to study abroad. IELTS if you’re interested in the UK. GATE if you want to study in India. I’ve seen friends turn paranoid before giving GRE and TOEFL. Bad GRE scores mean the admissions committee won’t even look at your application. But once you’re past a cut-off, it doesn’t matter at all. The rule of thumb is 90th percentile in quantitative analysis and >150/170 in the verbal section. It’s super easy to score well in TOEFL (>100/120). With good preparation, voracious readers can kick some serious ass in the GRE verbal section. >4.0 in AWA section of GRE is a requirement for a few universities.

How useful is having work experience before applying for highers? Is one year time for work experience ample enough?

Depends on the relevance of your work experience to the post-grad programme. It could be a plus for non-thesis/course based programmes. A research position/internship would be more valuable to a thesis-based masters or a PhD. However, most of my colleagues had >2 years of relevant work experience.

Your research focus was on Reinforcement Learning for Robotics. What kind of employment opportunities can one look forward to in Robotics?

Robotics is a burgeoning field. There’s been huge developments and significant investments in the field in the last couple of years. I’ve always been interested in the coding aspect of robotics. I do not build any robots. I rely on strong mechanical and electrical engineers to build robust, multi-sensory systems. I’m responsible for writing code that can give rise to useful, interesting behaviours. Robotics is an interdisciplinary research area that has something to offer for people in almost all areas of engineering. Just to give an example, I work with experts in deep reinforcement learning, classical control, motion planning, mechatronics, mechanical engineering and material sciences. The best location for jobs in robotics are Boston and Pittsburgh. They are home to a staggeringly large number of stalwarts and tech giants in the field of robotics.

How much did being in a tech club like Spider help you later in your Master’s degree major? What advice do you have for students who are not in tech clubs and yet are interested in pursuing a career in that field?

I was initially rejected by Spider. My closest friends and project collaborators (I worked with the same set of people on most of my projects) were rejected by RMI. None of us took it seriously in our first year. Once we worked on a few projects in the second year, 3 of us were inducted into Spider and RMI via lateral entry. Spider and RMI are great clubs to meet people with similar technical interests. That said, you can work on interesting projects on your own. No one is going to stop you from working on your interests. While I was not a part of Spider in my second year, I still used to regularly pester my seniors in Spider (Kowschick Boddu and Sandeep David) about embedded C coding. They were great mentors and helped me improve a lot. When I was working on the Texas Instruments Innovation Challenge 2014 (we were Phase 1 Winners and Finalists), my team had 2 not-so-active members of RMI, 1 non-tech club member (Aravind Govindan) and myself (a lateral entry in Spider).

My point is: tech clubs can be rewarding. I learned a lot from the Spider micro-controller workshops (once when I participated and once when I hosted) and had great respect for my teammates (especially PSS Srivignesh). Our team meetings were useful when we shared ideas. As long as you are working with the right people, it doesn’t matter if you are a part of a tech club. If you want to participate in international robotics challenges, funding is a game changer. Tech clubs can give you access to that. Arguably the best coder in my batch (Vignesh Sk) was not a part of any tech club. He’s now working with Walmart Labs in Bangalore. My good friends KM Ram Aravind and PR Siddharthan were not a part of any tech clubs and they did extraordinary work in Robotics (RMI was not even close!).

What courses helped you most in your area of research? Did any form of self-learning, like online courses, help you further your knowledge in that field?

While ICE dept is plagued by a large number of inept professors, they actually taught a lot of relevant courses. While the exams involved rote learning and hacky problem solving, the recommended textbooks and syllabus were useful to a certain extent. In hindsight, Digital Signal Processing, Neural Networks and Fuzzy Logic, Control Systems and Modern Control Theory were useful to me. Even though I didn’t learn much during the lectures, I had to look at these subjects during my internships and the recommended textbooks helped a lot. Some of the online courses and textbooks that helped me during undergrad:

  • Learning from Data by Abu Mostafa, Caltech
  • Machine Learning on Coursera by Andrew Ng
  • Autonomous Navigation for Flying Robots by TU Munich on edX
  • Convex Optimization from IIT Madras
  • Data Structures and Algorithms by CLRS
  • Signals and Systems on MIT Opencourseware by Alan Oppenheim
  • CS188 Intro to AI by UC Berkeley

You were also a web developer during your stint in NITT. Would you recommend branching out and experimenting by learning skills from various fields while in college?

To be honest, I hated being a web developer. I wanted to check it out since a lot of smart folks in Delta were doing it. I quickly found out that it wasn’t my cup of tea and modified my research interests. Like I mentioned earlier, college is a great place for exploration. Try something, fail fast and move on until you find your calling.

You did a fully funded thesis based programme at the University of Alberta, which is not easy to get. Do you have any advice on how to secure funding for such thesis-based masters programmes? Can you describe your experience as to how you got this programme?

I was surprised myself when I was offered funding. I don’t remember doing anything special and unfortunately, I do not have any advice about securing funding. Do good research, work on projects you’re interested in. With the right combination of luck and hard work, the right people would notice and good things happen! When I was working on my applications, I made sure that the universities I applied to had funding opportunities (in the form for Teaching and Research Assistantships (TA and RA respectively)) atleast after the first semester. For example, I didn’t even consider applying to the University of Southern California since they have near zero chance of funding for a Masters programme.

I had a GPA of 8.49 when I applied. Had I been a 9ptr, I’d have had a decent shot at Stanford and UC Berkeley. Except for myself, all my Indian batchmates at the University of Alberta were 9ptrs and batch toppers. In most universities, if you are outstanding in a course, you could TA for the same course in the subsequent year. Having a stellar academic record makes it more likely to secure funding. Getting an RA depends a lot on your professor. If you’re able to impress your supervisor and convince him/her that you are capable of fantastic research, he/she would offer an RA position depending on their funding situation.

You now work as an AI engineer at Kindred.Ai. What do you think is the scope of AI in India and abroad? How do those compare? Being a field that is steadily expanding, what opportunities do you think will come out of it in the future?

I can see a lot of interest in AI in India. I’ve been getting a lot of recruiter emails from India on Linkedin. But almost all those offers have something to do with Big Data Analytics. Amazon, Samsung, Microsoft Research, Yahoo Research, Nvidia, Xerox Labs, etc., are doing some amazing work in AI in India. That said, AI & Robotics research in India is still in the nascent stage. Companies like Grey Orange are doing some interesting stuff, but to the best of my knowledge, it’s not state-of-the-art AI. While the Indian government finally invested in a Ministry of AI, we’re way behind US, Canada and China. But I’m optimistic about India’s future in AI. We have a lot of talented people moving back to India and investing in local companies. Walmart buying Flipkart is a sign that Indian startups can make a successful exit. Hence, I foresee larger investments in the Indian startup ecosystem.

If the media hype about AI is even partially true, we would be heading into the second industrial revolution. A lot of people would lose their jobs and the transition would be hard and messy. The education system would need a massive overhaul and an ageing workforce would need retraining. In this future, coding would be an essential skill for survival. While manual labourers would be replaced by robots, we’d need more people in jobs which require interpersonal skills like geriatric care, human teachers for robots, etc. There’d be higher demand for creative and analytical minds. It opens up a new Pandora’s box called AI ethics and would require a lot of lawyers. It’d be interesting to see how societal forces (everyone ranging from trans-humanists to Luddites) react and shape the future.

What do you do now?

We’re working towards building machines with human-level intelligence. My main area of expertise is applying Reinforcement Learning and Deep Learning techniques to real-world robotics domains..