Analyzed 10M+ trade and tariff records using PySpark to perform distributed ETL and ML modeling to surface anomalous trends, generate summary statistics, and produce risk-oriented insights across 15+ sectors.

AI agent with FastAPI and Gemini that turns natural language goals into detailed project plans, creating Jira tickets, GitHub issues, and delivers automated stakeholder-ready reports in real time.

Fine-tuned Llama-3 8B using Supervised Fine-Tuning (SFT) on the Math QA Verification dataset (Hugging Face) to classify whether a given math answer is correct or incorrect based on the question and explanation context.

Agentic data analyst built with LangChain that turns natural‑language questions into CSV insights, visualizations, and summaries, enabling teams to make decisions without SQL or spreadsheets.

A fast and modular MCP-compatible server that intelligently scrapes web content and extracts structured, actionable insights in real time using LLMs like Gemini, GPT-4, and Claude.

Real-time yoga pose detection and intelligent classification using MediaPipe, Computer Vision, and Flask, designed to enhance wellness, mindful movement, and at-home fitness training.

A production-grade hybrid ML system that combines market data with real-time financial sentiment (from news, X, etc.) to predict short-term stock price movements for informed trading decisions.