B.Tech CSE Specializing in AI & ML, graduating 2027. I'm interested in the more mathematical and low-level side of deep learning — building models from scratch, understanding what's inside them, and working on problems where physics and neural networks intersect.
Currently learning CUDA and working on 3D Gaussian Splatting.
A ~60M parameter decoder-only transformer built entirely from scratch in PyTorch — no Hugging Face, no shortcuts. Implements RoPE, RMSNorm, SwiGLU FFN, Flash SDP attention with custom causal masking, and a per-layer KV cache with pre-allocated fixed-size tensors for zero-overhead inference. Trained on a weighted mix of FineWeb-Edu, Wikipedia, and SlimPajama with a custom SentencePiece tokenizer (32K vocab). Instruction-finetuned with response-only loss masking. Evaluated at perplexity 32.87 on a held-out split. Ships with a FastAPI streaming backend and a Svelte chat UI.
Physics-Informed GAN for Elastic Full Waveform Inversion — reconstructing subsurface Earth properties (Vp, Vs, density, Poisson's ratio, Young's modulus) from multi-component seismic waveforms. The generator is a U-Net that maps waveform inputs [B, 10, 1000, 70] to 70×70 subsurface grids; a Fourier Neural Operator acts as the differentiable elastic wave solver; a WGAN discriminator enforces realism. Total loss combines adversarial, data misfit (MSE), and PDE residual terms. Uses the ECFB dataset from the SMILE team. (In progress)
Agentic Scientific Research Platform. An agentic research system that investigates complex questions rather than simply generating summaries. Uses LangGraph multi-agent orchestration, hybrid BM25 + vector RAG, pgvector, Graphiti/Neo4j persistent research memory, evidence tracking, and reviewer-driven verification to produce structured, cited research reports. Designed around a core principle: LLMs handle reasoning; deterministic systems handle state, retrieval, evidence, and execution.
Vision-Transformer-for-DeepFake-Detection
ViT-based deepfake detector with self-supervised pretraining via masked image modeling on CelebA, finetuned on DFDC. Achieves ~85% accuracy and ROC-AUC ~0.93. Includes Grad-CAM to validate that detections focus on manipulated facial regions rather than background artifacts.
ESRGAN for 4× image super-resolution. Two-phase training: PSNR-optimised first, then adversarial + VGG perceptual loss. RRDBNet with 23 RRDB blocks. Tile-based inference reduces peak GPU memory from ~4.5 GB to ~500 MB (≈89% reduction) without degrading output quality.
Neural style transfer using VGG19 with multi-scale pyramid optimisation and L-BFGS refinement, balancing content, style, and total variation loss.
- Learning CUDA — kernels, memory hierarchies, warp-level operations
- Exploring 3D Gaussian Splatting for real-time radiance field rendering
Python · PyTorch · CUDA · C++ · FastAPI
Computer Vision · Language Modelling · Physics-Informed Neural Networks · Fourier Neural Operators · GANs · Self-Supervised Learning · Flash Attention · KV Cache · SentencePiece