Projects
Elixir Chess Engine
A 3500 ELO rated chess engine built from scratch in C++, with a hand-crafted evaluation, a custom-trained NNUE neural network, and a NegaMax search.
Birdly v2
A ConvNeXt-Tiny classifier for 236 bird species, with CPU and GPU runtime benchmarks, a FastAPI service on Cloud Run, and a Next.js frontend.
Adaptive Resampling based Training for Imbalanced Classification
A modality-agnostic adaptive resampling method to handle class imbalance in supervised classification. Paper submitted to MLWA Journal.
Elixir Chess Engine
A 3500 ELO rated chess engine built from scratch in C++, with a hand-crafted evaluation, a custom-trained NNUE neural network, and a NegaMax search.
xU-NetFullSharp Chest XRay Bone Shadow Suppression
Research Paper Implementation of the xU-NetFullSharp model for Chest X-Ray Bone Shadow Suppression.
Bird Image Semantic Segmentation
A bird image segmentation model using U-Net, trained on the CUB-200-2011 dataset.
DnCNN Image Gaussian Denoising
Research Paper Implementation of the DnCNN model for Image Gaussian Denoising.
India eBird Observations: EDA
Descriptive analysis of 30 million bird sightings in India, focusing on population trends of endangered species like the Pallas' Fish Eagle.
Birdly Bird Identifier
A large-scale bird classifier trained on 1M+ images of 1250 Indian bird species, using Google Cloud TPUs and a custom dataset.