Pinned Loading

1. Flash-Atten2-MMA Flash-Atten2-MMA Public

   FlashAttention-2 implementation from scratch. Features custom PyTorch C++ bindings, dynamic kernel dispatching, and hardware-accelerated Tensor Core MMA optimizations for D=128 LLM architectures.

   Cuda 1

2. black-scholes black-scholes Public

   Interactive, from‑scratch implementation of the Black‑Scholes model with full Greeks, P&L scenarios and a live Plotly dashboard. No external pricing libraries (QuantLib, etc.) are used – all formul…

   Python 1

3. Agentic-RAG-for-Time-Series-Analysis Agentic-RAG-for-Time-Series-Analysis Public

   An end-to-end multi-agent Retrieval-Augmented Generation (RAG) system built to provide a conversational interface for complex time-series data. This architecture seamlessly merges traditional data …

   Python

4. earning-predictor earning-predictor Public

   An AI-powered Financial Pipeline & Dashboard predicting whether a stock will BEAT or MISS its quarterly earnings.

   Python 1

5. Custom-HGEMM Custom-HGEMM Public

   A production-grade, highly-optimized Half-Precision General Matrix Multiplication (HGEMM) CUDA kernel built from scratch. This project demonstrates mastery over NVIDIA GPU architectures, shared mem…

   Cuda 1

6. cuda-insight cuda-insight Public

   CUDA Insight is an all-in-one educational toolkit and profiler designed to demystify GPU performance. Whether you're learning CUDA or optimizing a production kernel, CUDA Insight wraps NVIDIA's pow…

   Python 1