Building production-grade AI systems —
LangGraph agents, RAG pipelines, Kafka/Spark infrastructure.
Things that actually ship and actually work.
Full Stack Python Developer & AI Engineer building things that go beyond demos.
My flagship project is a LangGraph-powered GitHub PR Review Agent — Groq + LLaMA 3.1 70B, pgvector RAG, GitHub webhooks, deployed live on GCP Cloud Run. It reviews code like a senior engineer, automatically.
I also build real-time data pipelines with Kafka, Spark and Airflow, and full-stack apps with FastAPI backends and React frontends. I care about production-ready, not demo-ready.
AI agent that reviews GitHub PRs like a senior engineer — triggered via webhooks, fetches diffs, retrieves codebase context via RAG, posts structured inline comments.
Production-grade API monitoring and alerting with FastAPI, React, Firebase auth, Gemini AI diagnostics, and real-time uptime metrics.
GitHub →
End-to-end streaming pipeline — Kafka producer → Spark processing → ML inference → Airflow retraining → live dashboard.
GitHub →
Full-stack loan platform with JWT auth, ML credit risk scoring, admin dashboard, real-time installment calculation.
GitHub →
CNN-powered Flask app classifying MRI scans as Tumor / No Tumor with OpenCV preprocessing and MongoDB Atlas.
GitHub →
Fine-tuned ResNet-34 on plant disease classification with PyTorch. Real-time leaf image prediction through a Flask app.
GitHub →
Chat with any PDF via LangChain + vector search. Extracts, chunks, embeds and answers context-aware queries.
GitHub →Full Stack Python Developer and AI Engineer with hands-on experience building production-grade AI agents, RAG systems, real-time data pipelines, and full-stack web applications. Skilled in LangGraph, FastAPI, Kafka, Spark, and GCP. Delivered a live LangGraph + Groq PR Review Agent on Google Cloud Run.
Three.js physics simulation. Add particles, watch them orbit and collide.
Let's build
something real.
Open to AI/ML, backend, or data engineering roles. Also happy to collaborate on interesting projects. I reply within 24 hours.