PortfolioOpen to Work

Projects

AI and ML projects with real technical depth.

These projects reflect my strongest strengths: agent orchestration, RAG systems, ML systems, AI product design, and practical full-stack delivery.

Autonomous AI Operations01

Incident Response MCP System

An autonomous incident investigation platform built with a TypeScript MCP server, Python, LangGraph, AWS Bedrock, Pinecone, and React for incident analysis and postmortem generation.

Eliminated manual incident investigation across 85 plus incidents and 140 plus failure patterns through a 19-tool MCP system and self-improving multi-model RAG pipeline.
  • TypeScript
  • Python
  • LangGraph
  • AWS Bedrock
  • Pinecone
  • React
View GitHub
Applied AI02

PA Appeal Evidence Assistant

An agentic RAG product using Google Gemini 2.5, Voyage AI embeddings, Redis semantic caching, Supabase PostgreSQL, and a React interface for evidence retrieval workflows.

Demonstrated practical retrieval orchestration, caching, structured evidence assembly, and a strong AI product workflow.
  • Voyage AI
  • Supabase
  • Redis
  • FastAPI
  • Google Gemini
  • React
View GitHub
Financial Intelligence03

Financial Earnings Intelligence Agent

A multi-agent LangGraph system fine-tuned on Mistral-7B with QLoRA, streaming live financial insights through Redis pub/sub into a React and Recharts dashboard.

Delivered real-time market research and risk assessment insights through a fine-tuned multi-agent financial intelligence workflow.
  • Mistral-7B
  • QLoRA
  • Redis
  • LangGraph
  • React
  • Recharts
View GitHub
Research AI04

Intelligent Research Architect

An AI-powered research assistant for literature review, paper discovery, and document-aware research workflows built with FastAPI, LangGraph, Gemini, and Chroma.

Searches academic sources, supports PDF, text, and image uploads, maintains per-chat memory, streams source-backed answers, and exports results as Markdown and BibTeX.
  • Python
  • FastAPI
  • LangGraph
  • LangChain
  • Google Gemini
  • Chroma
  • Tavily
  • React
  • TypeScript
  • Vite
  • Tailwind CSS
View GitHub
Document AI05

DOCUAI

A serverless document question-answering system built with AWS Lambda, Amazon Textract, and AWS Bedrock using Llama 3 70B Instruct.

Extracts text from uploaded PDFs or images, answers custom questions about the document, and uses API Gateway, IAM, and S3 for secure serverless deployment.
  • Python
  • AWS Lambda
  • Amazon Textract
  • AWS Bedrock
  • API Gateway
  • S3
  • IAM
View GitHub
Computer Vision06

Sign Language Recognition

A real-time sign language recognition system built with convolutional neural networks and OpenCV for gesture translation from webcam input.

Recognizes sign gestures in real time using a custom CNN model trained on the Sign Language MNIST and ASL dataset for intuitive communication assistance.
  • Python
  • TensorFlow/Keras
  • OpenCV
  • NumPy
View GitHub
Face Recognition07

Smart Attendance

A computer vision-based smart attendance system that uses facial recognition to automatically mark attendance in real time.

Trains custom face datasets for identification and automates attendance logging for classrooms, offices, and events using OpenCV and Python.
  • Python
  • OpenCV
  • face_recognition
  • NumPy
  • Pandas
View GitHub