Welcome! Thanks for taking an interest in my work. I am Sayantan. I am an Astrophysicist at NASA-JPL. I have an amazing job where I spend my time thinking about the workings of the cosmos. I enjoy the opportunity to solve research problems that help us to have better insights into the formation and evolution of stars and planets. I use supercomputers to run sophisticated numerical simulations to model the interstellar systems to interpret observed data from large Telescopes. Additionally, with the aid of Machine Learning algorithms, we train models to understand and interpret data and bridge a bond between simulated models and real observed data.
As a scientist, my main focus in research revolves around addressing fundamental inquiries about natural phenomena, such as the origin of star-planet systems, through the utilization of cutting-edge numerical simulations. My efforts are directed at constructing models that elucidate observational data and unravel the inherent physics. To achieve this, I employ a hybrid approach that combines physics-inspired theoretical models with computer simulations, complemented by advanced Machine Learning (ML) techniques. My vision is to progress toward my research objectives by constructing robust physics-informed ML models that harness the capabilities of the next generation of principled ML and optimization algorithms.
​
My profession allows me to travel around the globe attending scientific meetings and conferences to present my work and exchange ideas. As an ardent traveler, I often use this opportunity to see new places and enjoy the local culture and cuisine.
​
When I am not working, I love spending time with my family. I do love sports, especially cricket. I also enjoy playing racquet sports like badminton and table tennis.
RESEARCH INTERESTS
-
Physics-informed neural networks, Transformer PINNs
-
Generative Algorithms like GAN, Diffusion models
-
Computer vision using CNNs
-
AI in Magnetohydrodynamics and Computational Astrophysics
​
Work Experience
Aug 2022- Present
Nov 2020 - June 2022
Nov 2018 - Oct 2020
May 2017 - Oct 2018
Computational Astrophysicist | Machine Learning Scientist: NASA-Jet Propulsion Laboratory
​
I design deep learning models using Physics informed neural networks (PINNs) to solve PDEs. I developed GRINN, PINN-based code, to solve coupled time-dependent PDEs to simulate complex hydrodynamic systems. Obtained 10 times more efficient GRINN code compared to the Finite difference scheme.
​
Develop a conditional Generative Adversarial Network (GAN) based model to augment (perform rotation, radiative transfer) scientific images. Adapting cGAN to generate new images (saving 100s of hours of computation compared to conventional methods)
Postdoctoral Researcher in Data-Intensive Astronomy: Iowa State University
​
Designed Bayesian Neural Network (BNN) model using Bayesian inference (“DPNNet-Bayesian”) to better understand and quantify the uncertainty associated with deep neural network predictions. Understanding epistemic and aleatoric uncertainties.
Postdoctoral Researcher | Computational Astrophysicist: Institute of Astronomy and Astrophysics
​
Conceptualized, designed, and developed “Disk Planet Neural Network" (DPNNet) using Convolutional Neural Networks (CNN) to predict the mass of unseen exoplanets (saving 100s of hours of computation time) from images observed using the ALMA Telescope.
Research Fellow: Harvard University - Center for Astrophysics
​
​Interpretation and statistical analysis of complex telescope data using theoretically motivated models resulting in journal publications.
Education
Sep 2014 - Aug 2018 Ph.D. Physics
Sep 2013 - Aug 2014 M.Sc. Astronomy
Jun 2011 - June 2013 M.Sc. Physics
Jun 2008 - June 2011 B.Sc. Physics
The University of Western Ontario
​
The University of Western Ontario
​
Thesis: “A Study of Substellar Mass Limit for Brown Dwarfs”
Indian Institute of Technology, Madras
​
Thesis: “Inflation and Cosmological Perturbation Theory”