Hello, I'm Jagan
I am a passionate Full-stack AI Software Developer specializing in LLM integration and retrieval optimization in modern, responsive, and user-friendly web applications. Let's create something amazing together!
About Me
I am a Fullstack AI Engineer with a passion for creating interactive and responsive web applications with AI and LLM integration. I have experience working with JavaScript, TypeScript, React.js, Next.js, Node.js, PostgreSQL, and Python. My recent project: rag-document-parser — a modular RAG toolkit with semantic + lexical search (llama-text-embed-v2, multilingual-e5-large, BM25, Pinecone sparse), rigorous evaluation (Precision@k, Recall@k, Hit Rate) tracked in MLflow, hybrid retrieval & reranking (bge-reranker-v2-m3, cross-encoder/ms-marco-MiniLM-L-12-v2), and production observability (Prometheus, Grafana, OpenTelemetry) documented via ADRs.
- Python
- Node.js
- React.js
- PostgreSQL
- TypeScript
- JavaScript
- Ruby on Rails
- Java
My Projects
Favourite Stocks Dashboard
Built a Fullstack dashboard on Vercel using Next.js, providing real-time data from Polygon.io APIs. Implemented dynamic routing for detailed company pages, persistent data with Neon's serverless Postgres, and automated nightly data updates using Vercel's cron jobs.
Gemini AI Chatbot
A simplified AI project built with Next.js. This side project helped me to quickly experiment with AI integrations and learn the basics of building AI-powered applications. Built with Next.js, AI SDK, Tailwind CSS, TypeScript, and shadcn, it provides a minimalistic foundation for creating AI chatbots, focusing on simplicity and speed.
Projects Board
A collaborative project management board built with Next.js and TypeScript. This tool allows teams to organize, track, and manage tasks efficiently with a modern UI and real-time updates. Features include drag-and-drop task management, user authentication, and integration with popular developer tools.
Projects Board Tenant Server
A multi-tenant backend server for the Projects Board application. Built with Ruby on Rails, it provides secure APIs, tenant isolation, and scalable architecture for managing multiple organizations and users. Designed for reliability and easy integration with the Projects Board frontend.
Projects Board App
A mobile-optimized Kanban-style project management board built with Expo React Native. Features include touch-friendly task management with To Do, In Progress, and Done columns; secure JWT authentication with multi-tenant support; create, edit, delete, and move tasks; native iOS design with rounded corners and smooth animations; offline-ready with AsyncStorage; task priority and due date management.
RAG Document Parser
A comprehensive Retrieval-Augmented Generation system built as a Full-stack AI Engineer, featuring Python, Flask, Pinecone vector database, and advanced ML monitoring. Architected modular ingestion pipeline with PyPDF2 for document processing, semantic & lexical search capabilities, intelligent chunking, and metadata extraction. Implemented hybrid retrieval combining dense+sparse search with bge-reranker-v2-m3 for reranking, achieving 50% improvement in retrieval precision (Precision@5 from 0.400 to 0.600). Integrated real-time monitoring with Prometheus & Grafana, experiment tracking with MLflow, observability via OpenTelemetry, and Google Gemini for enhanced generation. Documented architecture decisions with ADRs for maintainability and team collaboration.
My Work Style
Using VSCode & GitHub Copilot
I leverage the power of modern tools like VSCode and GitHub Copilot to enhance my productivity and streamline the development process. These tools help me write clean, efficient, and maintainable code while reducing repetitive tasks.
With GitHub Copilot, I can quickly generate code snippets, debug issues, and explore new ideas, making programming faster and more enjoyable. Combined with the flexibility of VSCode, I can focus on solving complex problems and delivering high-quality solutions.
