About Me
Researcher. Builder. Perpetual explorer of ideas at the edge of what is known.
I'm a PhD candidate in Computer Science at BITS Pilani, Goa, specializing in computer vision and foundation models. I'm currently a Research Fellow at BITS Pilani, K K Birla Goa Campus, where I work on industry-sponsored ML systems for automotive lifecycle and safety.
My work spans NLP, robotics, biomedical signal processing, and especially Computer Vision. I closely follow foundation model research across vision and language, and I'm increasingly drawn to 3D vision and its open problems. Healthcare is a space I care deeply about.
I love to explore interesting things, innovate, and contribute to knowledge for the future of humankind. We can talk about barmy things to invent, build, engineer. Not limited to but we can also talk about literature, philosophy, economics, sociology, geopolitics, et cetera :)
I've previously interned at IISER Kolkata, Defence Institute of Advanced Technology (DIAT-DRDO), and IIT Bombay, and co-founded CodePVG during my bachelor days, a campus tech community that brought together 1000+ students.
Education
Academic foundations in computer science and engineering
Experience
Research and engineering roles at premier institutions
Research Fellow
BITS Pilani, K K Birla Goa Campus
- ▹Working on an industry-sponsored ML project for end-to-end automotive lifecycle and safety systems.
- ▹Developing and evaluating ML models and pipelines for real-world, large-scale data.
Research Intern
Indian Institute of Education Science and Research (IISER), Kolkata
- ▹Developed SportsOpi, a 43K-comment dataset on six major sports controversies, using LLaMA-3.1's multi-phase annotation and human verification (κ > 0.8) for reliable 4-way stance labels.
- ▹Fine-tuned LLaMA-3.1-8b and DeepSeek-8b, raising stance-detection F1 from ~0.35 to ~0.75 and performed attention-based probing to interpret model gains.
Research Intern
Defence Institute of Advanced Technology, Defence Research and Development Organisation (DIAT, DRDO)
- ▹Developed a novel EEG signal classification method for upper limb movement tasks, achieving 97.50% micro-averaged accuracy and 99.17% mean accuracy, outperforming recent approaches.
- ▹Extracted and evaluated time-domain statistical and band-power features from pre-processed EEG signals using multiple machine learning classifiers, including KNN, SVM, and an ensemble classifier, improving motor imagery task classification accuracy by 10% with MATLAB and Python.
Project Intern
Indian Institute of Technology (IIT), Bombay
- ▹Project Title: Brain Tumor Detection. Implemented a Convolutional Neural Network (CNN) for Brain Tumor Detection which achieved a robust testing accuracy of 95%, demonstrating superior generalization ability in identifying brain tumors from MRI images.
- ▹Utilized a pre-trained ResNet-50 model for brain tumor detection that achieved a high training accuracy of 99.50%.
- ▹Conducted a comprehensive comparative analysis between CNN and ResNet-50 models.
Projects
Research work and engineering projects that push boundaries
Parameter-Efficient Automotive Damage Detection via Deformable Spatial Adapters
Bridging rigid foundation model features to irregular dent geometry using DCNv4 adapters.
Robust by Design: Benchmarking Hyperdimensional Computing Against Neural Baselines Under Hardware Faults
HDC stays above 94% accuracy where FNNs and Transformers collapse to near-random at sub-0.1% bit-flip rates.
TPACT: Transformer Positional Encoding with Adaptive Coordinate Tokenization for 3D Point Clouds
A geometry-aware positional encoding method that outperforms existing PE variants on real-world noisy point cloud data.
Retinal Micro-Lesion Detection via RETFound-DETR: Fine-Tuning a Foundation Model for Diabetic Retinopathy Screening
Adapting a retinal foundation model (ViT-Large, pretrained on 1.6M fundus images) for bounding-box-level micro-lesion detection.
Comparative Study of Detection and Segmentation Architectures for Oral Cancer Lesion Analysis
Evaluating Mask R-CNN, UNet++, and RTMDet for oral cancer detection and segmentation.
SportsOpi Dataset & Stance Detection
EEG Signal Classification
AI-Powered Spirometer
Brain Tumor Detection
EMG-Based Robotics Claw (Prosthetics)
Skills
Technical proficiency across ML, development, and cloud
Core Competencies
Leadership & Volunteering
Building communities and driving impact beyond research
General Secretary
TESA (E&TC Student's Association), PVGCOET
Aug 2023 - Aug 2024
- ▹Led flagship TECHCRAFT national tech event (20+ yr legacy)
- ▹Brought major sponsorships from industry
- ▹Organized seminars with guests from top institutions
Co-founder & Networking Head
CodePVG
Jan 2022 - Apr 2025
- ▹Produced podcasts with IIT Bombay alumni, Goldman Sachs NY VP, and author of CS books
- ▹Hosted sessions on Python, Tinkercad, LinkedIn
- ▹Built campus peer learning ecosystem for 1000+ students
Event Organizer
Google Developer Student Clubs (GDSC)
Aug 2022 - Aug 2023
- ▹Organized React/GitHub workshops
- ▹Co-managed Annual Ideathon (multi-college)
- ▹Logistics and permissions for large-scale events
Get in Touch
Let's collaborate on impactful research and projects