Projects
The first cross-layer mapping framework between submarine cables and IP links at a global scale. Built entirely on public data sources, Nautilus has an impressive mapping accuracy on validation.
Nautilus: A Framework for Cross-Layer Cartography of Submarine Cables and IP Links
Sept 2023
Network Flow Prediction and Control
Dec 2022
Based on the Flux project, this work involved an analysis with multiple Machine Learning models for Network Flow prediction. This work also explored the applicability of a single ML for the flow prediction problem.
In this project we leverage Deep Convolution Neural Networks and Transfer Learning to predict Traumatic Brain Injury (TBI) in humans and mice based on EEG data
Transfer Learning on EEG Data
Jun 2022
Sokoban Solver with Reinforcement Learning
Dec 2021
This project involved the application of Reinforcement Learning to solve the Sokoban game. For efficiency, A3C model with action masking was employed.
A Computer Vision project that leveraged LIDAR and cameras to detect and classify objects on the road. This model leveraged the individual merits of each sensor to develop a better occlusion detection mechanism and learn an accurate model.
Multi-sensor-based detection and classification of road obstacles
Jun 2018