Physics-guided deep learning for image reconstruction and inverse problems
I'm a PhD researcher at the Biomedical and Astronomical Signal Processing (BASP) Laboratory, Heriot-Watt University in Edinburgh, working on physics-based image reconstruction for radio astronomy, including imaging black holes. I came to research from three years as a software developer at TCS, then a triple Master's in Image Processing and Computer Vision across Hungary, Spain and France (4.51/5, second in my batch). I like applied maths, inverse problems and fast, GPU-accelerated code.
PyTorch
TensorFlow
Keras
Scikit-Learn
NumPy
Pandas
Matplotlib
Astropy
OpenCV
Scikit-Image
Pillow
Python
MATLAB
C++
SQL (MS-SQL)
TeX
Git
LaTeX
Jupyter Notebooks
Anaconda
Visual Studio Code
Google Colab
Kolkata, IndiaB.Tech, Amity · TCS2015 – 2022
Budapest, HungaryMSc, PPCU2022
Madrid, SpainMSc, UAM2023
Bordeaux, FranceMSc + thesis, UBx / IMS2023 – 2024
Edinburgh, UKPhD, Heriot-Watt2024 – nowStarted my research journey at BASP lab in Heriot-Watt University, UK
Working on physically guided deep radio-interferometric image reconstruction with deep learning
Commenced my master thesis at the IMS Laboratory in Bordeaux, France
Worked on parameter tuning in tomography to develop an optimized and guaranteed algorithm
Commenced my third semester of the Master's program at the University of Bordeaux, France
Delving deeper into applied mathematical approaches to AI and Computer Vision
Began my second semester of the Master's program at the Autonomous University of Madrid, Spain
Explored various applications and theories behind deep vision models, which greatly piqued my interest
Initiated my second internship (online) at the Indian Statistical Institute, Kolkata
Gained insights into various models and the mathematics underpinning deep learning
Embarked on my first semester of the Master's program at Pazmany Peter Catholic University, Budapest, Hungary
First of three semesters across Hungary, Spain and France in the IPCV Master's programme
Awarded a 100% scholarship for pursuing a Master's in Image Processing and Computer Vision
Left my software engineering role at TCS to pursue the programme full-time
Started my career at TCS as an Assistant Systems Engineer trainee
Frontend development with React JS, plus SQL databases and some .NET (C#) backend work
Completed my first internship at the Indian Statistical Institute, Kolkata
Ventured into the realms of computer vision, image processing, and machine learning
Commenced my undergraduate (B.Tech) studies in Electronics and Communication Engineering at Amity University, Kolkata
Initially aspired to be an electronics engineer, working with integrated circuits (ICs)
This is my masters thesis project on tomographic imaging and parameter learning using the famous Shepp-Logan phantom.
This project deals with the segmentation of optic disc from retinal fundus images from the famous Drishti-GS dataset
Dice 93.95% on Drishti-GS (published 2025)
3-class classification of cosmological objects in the famous SDSS dataset, version DR14
99.4% accuracy with XGBoost
Journal of Neonatal Surgery, Vol. 14, No. 15S, 2025
A fast, lightweight method for segmenting the optic disc in retinal fundus images using thresholding and morphological operations, evaluated on the public Drishti-GS dataset.
Dice 93.95% · Jaccard 88.9% · centre-localisation accuracy 99.67%
Selected from among top-tier candidates as the recipient of the prestigious James Watt Scholarship for PhD research at Heriot-Watt University.
Selected among Synapse AI Symposium’s top 50 candidates at Milan, alongside global AI luminaries.
Secured Cybathlon 2023-24’s Vision Assistance Challenge victory among 21 global teams at ETH Zurich.
Chosen along with 9 applicants to receive the prestigious Erasmus+ Grant as funding for studies in Europe.
Granted full IPCV scholarship as one of only 3 non-EU citizens selected from more than 1000 applicants.
Earned Silver Medal for B.Tech ECE with 9.26/10 GPA across 4 years.