Projects
Experience

Jan '22 - Present
Coming up with a structured way to compare and contrast representations produced by various domain generalization algorithms in the field of computer vision. Please reach out for a copy of our findings!

Aug '21 - Present
Formalizing the modality selection optimization problem via tools from submodular maximization and information theory. Work accepted at UAI 2022 - camera ready coming soon!

Sept '21 - Dec '21
Worked on the problem of detecting image manipulation, taking inspiration from the NeurIPS 2021 competition on Image Similarity.
Jan '20 - Jun '20
Worked on developing a novel method for producing prediction intervals in regression, inspired by model callibration.

May '19 - Aug '19
Introduced the idea of using ILP for obtaining comprehensible cyber-physical system controller strategies while incorporating domain knowledge. Integrated a strategy visualisation feature in the PRISM Model Checker. Solo authored publication at ASE 2019. ASE 2019 SRC Undergraduate Third Prize winner.

May '18 - Aug '18
Was part of the backend development team at the Gnowledge Lab led by Nagarjuna G.. My team and I introduced the capability to export the course content for offline viewing in the form of an epub. We introduced cross-platform export capabilities with Moodle, a widely known e-learning system.
Academic
Studied Deep Probabilistic logic (DeepProbLog) to leverage the representational power of neural networks for high level logical reasoning.
Presented work in talks at TCS Research, Delhi and TCS Research, Kolkata.
Applied machine learning techniques on channel state information (CSI) captured by the ESP32 IoT device to predict useful information for movement detection, ultimately aiming to provide indoor localisation.
Demonstration paper accepted at the International Conference on COMSNETS 2020.
Worked with the 2000 Qubit D-Wave Quantum Annealer hosted by USRA, NASA to efficiently solve classically intractable QUBO optimisation problems like SAT. Furthermore, explored the possibility of such a speed-up for Boltzmann Machines.
Talk given at the International Quantum Computing Symposium colocated with ACM ICPE 2019, IIT Bombay.
Designed and implemented a parser for processing transcripts and invigilation sheets for the Academic Under Graduate Studies Division (AUGSD). Designed in Python, Elixir.
- Modified WalkSAT Implementation
- 8 Puzzle using A* search
- Multimodal Transport Scheduler
- TicTacToe Bot
- Quantum Machine Learning
- Airport Management System