Gautam Pai

Postdoctoral Researcher, Centre for Analysis, Scientific Computing and Applications, Eindhoven University of Technology

Gautam Pai is currently a postdoctoral researcher in the Mathematics and Computer Science department at TU/e. He did his Ph.D. in the Faculty of Computer Science at the Technion – Israel Institute of Technology, following which he spent a couple of years as a postdoctoral researcher at Ecole Polytechnique in Paris. He also spent a couple of years working as an engineer with Philips Research Bangalore, in India.

Gautam’s research interests broadly span different aspects of geometry in computer vision and machine learning. In particular, he is interested in developing computational methods for exploring and understanding geometric structures in data. He works on the applications of spectral and distance geometric methods to geometry processing, 3D shape analysis, unsupervised learning, and image processing.

Gautam’s career is aided by some generous fellowships from the Technion Cyber Security Research Center and the Lady Davis fellowship trust. He is also the recipient of a couple of best paper awards at leading international conferences.

3 questions to Gautam Pai

Having been affiliated with three EuroTech Universities, how would you summarize your overall experience at these institutions?

Fantastic. I am truly privileged to work closely with some of the best research groups in my field. My Ph.D. studies at the Technion helped me establish a solid foundation in the basic tricks of the trade. Amongst many things, this included exposure to high-quality graduate courses and regular interactions with researchers working on congruent topics but spread across different departments. For example, I was a part of the Geometric Image Processing Laboratory, but I had regular fruitful interactions – by way of courses, seminars, and group meetings with members from the Center for Graphics and Geometric Computing, the Visual Sensing Theory and Applications Laboratory and the Geometry Oriented Signal Processing Laboratory from the electrical engineering and computer science departments. I especially valued the culture of my group which encouraged originality in research and promoted an open collaborative atmosphere. I think such an environment helps develop confidence in one’s worldview, especially for establishing a research agenda during a Ph.D.

During my postdoctoral role at Ecole Polytechnique, I observed and learned to establish the work ethic needed to execute successful research collaborations. When working in a team, a lot of work goes into converting a research idea into a research project, especially in the early stages when there are a lot of unknowns. Thanks to an amazing group leader and along with some very talented colleagues, one such project was recognized as the best paper award at a leading international conference last year. I specifically would like to mention that despite a very difficult period due to the pandemic, I have thoroughly enjoyed and benefited from weekly online group meetings and seminars at the Geometric Visual Computing group at the Polytechnique.

My current role at TU/e complements these previous positive experiences. I am excited to work with enthusiastic and motivated colleagues and a very comfortable environment both personally and professionally at the TU/e.


In hindsight, what was/were the decisive factor(s) that got you where you are today?

I tend to get my drive from curiosity. I know that there is also a strong career aspect to work-life – you need to publish good papers, get admitted to good places, get recognized, etc. However, I have observed that when times are tough: you get a paper rejected, or your experiment does not work or you discover a bug in your code one day before a deadline :), it helps to fall back on a basic emotion of curiosity. At least for me, things have eventually fallen into place with this mindset. It may not be exactly the route I intended, but a satisfactory path always shows up when I focus on the “why?” in the problem. I suppose this is the strongest factor that has kept me going. This philosophy was reinforced by the positive attitude of my advisors and colleagues who encouraged and aided my interests all through. In addition, I also try to practice a good ethic for communication – writing, verbal, and otherwise and I feel that I have benefitted from it. I think that communicating clearly and objectively structures and simplifies complex conversations and I found this useful to make progress.


What project or endeavour are you looking forward to?

My current research focus is the domain of Geometric Learning. The overall goal is to develop algorithms for efficiently processing data that has some underlying mathematical structure. More simply, this involves building software to process various kinds of shapes: from one-dimensional curves, or two-dimensional surfaces to higher dimensional geometries (manifolds), with important applications in the areas of image processing and computer vision.

The focus during the early stages of my graduate studies centered on geometric analysis that significantly relied on mathematical concepts for processing such data. In recent years, I’ve spent some time working on machine learning which I see as a data-driven methodology to build algorithms for various applications. However, it’s not immediately clear how to incorporate such mathematical constraints into these data-driven models and whether doing so is useful at all. Initially, it appeared that this is a binary – you either learn everything from data and have little interpretability or have maximum mathematical preciseness yet, be sort of underwhelming when it comes to the performance in the application. It is very exciting to discover problems that sort of blur this binary and use mathematical insights to build better, interpretable machine learning models. This is precisely the focus of the project I am currently affiliated with at the TU/e called “Geometric Learning for Image Analysis” and I am looking forward to hopefully many interesting discoveries in this project.

Share on facebook
Share on twitter
Share on linkedin
Share on email