An interview with Shalini De Mello, Principal Research Scientist at NVIDIA
Our interviewee for today, Shalini is a Principal Research Scientist at NVIDIA. Her research interests are in computer vision and machine learning for human-computer interaction and smart interfaces. At NVIDIA, she has invented technologies for gaze estimation, 2D and 3D head pose estimation, hand gesture recognition, face detection, video stabilization, and GPU-optimized libraries for mobile computer vision. Her research over the past several years has pushed the envelope of human-computer interaction in cars and has led to the development of NVIDIA’s innovative DriveIX product for smart AI-based automotive interfaces for future generations of cars.
Previously, she has worked as an Imaging and Architecture Scientist at Texas Instruments (2008–2010) designing algorithms for the image signal processing pipeline of mobile phones. Additionally, she has conducted research internships at AT&T Laboratories (2007), Advanced Digital Imaging Research, LLC (2004–2006), the Indian Institute of Science, Bangalore (2001), and the Indian Institute of Technology, New Delhi (2000).
Shalini received her Masters and Doctoral degrees in Electrical and Computer Engineering from the University of Texas at Austin in 2004 and 2008, respectively where she worked with Dr. Alan Bovik and Dr. Mia K. Markey. She received a Bachelor of Engineering degree in Electronics and Electrical Communication Engineering from Punjab Engineering College, India, in 2002. She is a recipient of the prestigious Summer Research Fellowship (2001) awarded by the Jawaharlal Nehru Center for Advanced Scientific Research, Bangalore, India.
I would like to wholeheartedly thank Shalini for taking the time to do this interview. I hope this interview serves a purpose towards the betterment of data science and machine learning communities in general :)
An interview with Shalini De Mello, Principal Research Scientist at NVIDIA
Sayak: Hi Shalini! Thank you for doing this interview. It’s a pleasure to have you here today.
Shalini: Thank you, Sayak, for having me. It is a pleasure to be here.
Sayak: Maybe you could start by introducing yourself — what is your current job and what are your responsibilities over there?
Shalini: I am a Principal Research Scientist at NVIDIA Research. I conduct fundamental research on designing machine learning and computer vision algorithms for human-computer interaction and smart interfaces, especially inside cars.
Sayak: Ah, that’s nice! Great to know that your work involves computer vision as mine does too. What motivated you to take machine learning as one of your areas of interest?
Shalini: Very early on in my undergraduate studies I did internships at IIT Delhi and IISC Bangalore. There I was introduced to the field of computer vision, which fascinated me. I was hooked and knew I wanted to make a career of doing research in that area. Computer vision (CV) and machine learning (ML) go hand-in-hand. Machine learning is the biggest tool that a computer vision scientist has today. It is THE tool to solve complex problems like vision and speech understanding and natural language processing that we don’t have analytical solutions for.
Sayak: That is so very true! When you were starting what kind of challenges did you face? How did you overcome them?
Shalini: When I started working in machine learning back in 2001 the biggest challenges were the lack of availability of sufficient computational power in the processors of the time to enable real-world ML/CV applications. I joined NVIDIA in 2011, as I felt that GPUs as accelerated computing devices presented the only feasible solution for deploying ML and CV algorithms in the real world.
The lack of data for training ML models was also a big issue back then, but the internet boom largely solved it.
Sayak: Yes, those 10 years changed many things not only in the machine learning world but also in the world of high-performance computing. What were some of the capstone projects you did during your formative years?
Shalini: There were two capstone projects that I did during my summer internships as an undergraduate student that have heavily influenced my career. The first was at IIT Delhi in the summer of 2000, where I was introduced to the field of image processing and I did various small projects to manipulate the appearance of images. The second project was at IISC Bangalore in 2001, where I built a face recognition system using the original Eigenfaces algorithm proposed by Turk and Pentland.
Sayak: Back in those early days, these projects must have been really fun! Fields like machine learning are rapidly evolving. How do you manage to keep track of the latest relevant happenings?
Shalini: I try to expose myself to as many sources of knowledge as I can. I constantly talk and discuss ideas with my peers and colleagues within NVIDIA Research and with external collaborators. I constantly read papers published by well-respected researchers in the field. I review papers for 7–8 top machine learning and computer vision conferences and attend 2–3 of them each year. I try to regularly attend invited talks by experts in their fields that happen internally within NVIDIA.
Sayak: That is quite a toll to keep yourself all updated, really! Being a practitioner, one thing that I often find myself struggling with is learning a new concept. Would you like to share how do you approach that process?
Shalini: Start at the basics and build your way up. That I feel is the only way to truly and fundamentally understand a concept.
Sayak: I really could not agree more with that! Many students find it very hard to properly do research in the field they are interested in. Would you like to share your thoughts on this?
Shalini: You have to have genuine curiosity and interest in what you are researching. If you are not truly interested in the problem you will quickly lose motivation. Also, don’t be afraid of failure. In research, there are many times when you have a certain intuition for how to solve a problem, but when you actually practically try it, it doesn’t work out. It is important to not have the expectation that every single idea that you try will work out as intended. Success is built upon many failures and hard work.
Sayak: I can relate to each and every word there. Thanks for sharing that, Shalini. Any advice for the beginners for being successful in machine learning?
Shalini: Learn the basic mathematical concepts and one state-of-the-art framework like PyTorch or TensorFlow. Then quickly start applying what you have learned to real-world problems. The more problems that you try and explore the more confidence you will get with practically using machine learning and develop your own intuition around how to use it for various new problems.
Sayak: Learn, implement and repeat! I have also found this three-steps’ principle to be highly effective. Thank you so much, Shalini, for doing this interview and for sharing your valuable insights. I hope they will be immensely helpful for the community.
Shalini: Thank you for having me. It was a real pleasure.
Summary
Shalini’s views and experiences are very inspiring. She went on her computer vision journey back in the days when owning a computer especially in countries like India used to be a big thing. From there, she is working as a Principal Research Scientist at an organization that is constantly pushing the boundaries of high-performance computing. Her strong willingness to learn is also something that we all should embrace. Lastly, her words on doing research with strong determination are very noteworthy.
I hope you enjoyed reading this interview. Watch out this space for the next one and I hope to see you soon. This is where you can find all the interviews done so far.
If you want to know more about me, check out my website.
