Full Stack Python Developer & AI/ML Engineer based in Delhi, India 🇮🇳

I build production-grade AI systems and data pipelines — from LLM-powered agents to real-time streaming infrastructure. My work sits at the intersection of backend engineering, machine learning, and agentic AI.

• 🔭 Currently building CODEREVIEW-AI-AGENT — a LangGraph stateful agent that reviews GitHub PRs using Groq + LLaMA 3.1 70B, deployed on GCP Cloud Run
• 🤖 Experienced in RAG systems, LLM integration, and multi-step AI agents using LangGraph & LangChain
• ⚙️ Strong in data engineering — real-time pipelines with Kafka, Spark, and Airflow
• 🌐 Full stack proficiency: FastAPI / Flask backends + React frontends + PostgreSQL / MongoDB databases
• 🚀 Comfortable with end-to-end deployment: Docker + GCP Cloud Run
• 🏆 Hackathon Winner @ TechFest 2024

📬 Reach me: pushstack25@gmail.com · LinkedIn

An AI agent that reviews Pull Requests like a senior engineer

• LangGraph stateful agent: fetch → analyze → review → post
• Groq + LLaMA 3.1 70B for blazing-fast code understanding
• pgvector RAG for context-aware, codebase-aware reviews
• GitHub Webhooks → auto-triggers on every PR
• FastAPI + async SQLAlchemy backend with full persistence
• Deployed on GCP Cloud Run via Docker

Python LangGraph Groq pgvector FastAPI PostgreSQL Docker GCP

End-to-end streaming data pipeline predicting NYC taxi demand in real time

• Kafka producer → Spark Streaming consumer → feature engineering → ML inference
• Airflow scheduled retraining at fixed intervals
• Dash dashboard + PostgreSQL for historical storage

Kafka Spark Airflow PostgreSQL ML Dash Python

Full-stack loan management + ML-based risk prediction platform

• JWT auth with user/admin roles
• Real-time installment & interest calculation
• ML risk scoring integrated into loan approval flow
• FastAPI backend + React frontend + PostgreSQL

FastAPI React PostgreSQL ML JWT Python JavaScript

CNN-powered MRI classification web app

• Pre-trained CNN model classifying Tumor / No Tumor
• Flask web app with file upload + MongoDB Atlas storage
• Image preprocessing with OpenCV + Keras

Python CNN Keras Flask OpenCV MongoDB Atlas

Real-time plant disease detection with fine-tuned ResNet-34

• Fine-tuned ResNet-34 on plant disease dataset using PyTorch
• Flask web app with live image classification

PyTorch ResNet-34 Flask Computer Vision

Chat with any PDF using LLM + vector search

• PDF text extraction → chunking → embedding → vector DB retrieval
• Context-aware Q&A via LangChain + LLM

Python LangChain RAG Vector DB NLP

🥇 Hackathon Winner @ TechFest 2024
📦 Published Open Source NPM Package
🤖 Built production LangGraph AI agent deployed on GCP Cloud Run
🧠 End-to-end real-time ML pipeline with Kafka + Spark + Airflow

"Ship fast. Learn faster. Build things that matter."