Featured Work
Projects
Several projects that I have worked on, both private and open source.
Agentic workflow application that uses LLM-powered agents to break down tasks into steps and execute AI-driven workflows. Recognized as a Top 10 project at the Luna Hackathon.
Autonomous AI email pipeline with IMAP integration that categorizes incoming emails into spam & sales leads, generates AI lead scores based on responses, and includes a human-in-the-loop approval procedure to review, edit, and dispatch automated emails.
Minimalist, Japanese-inspired web UI component design system focused on extreme clarity, typography precision, elegant whitespace, and neutral color palette.
Modern, rich-aesthetic SaaS landing page and product dashboard interface design featuring glassmorphism, responsive grid architecture, and interactive components.
AI-Powered comprehensive study platform combining intelligent learning paths, interactive PDF chat, and AI-driven resource curation.
A comprehensive crowdfunding platform built with React.js featuring user registration, project creation, pledge tracking, and milestone funding mechanisms.
An intelligent chat application powered by AI agents featuring real-time conversations, smart agent handoffs, and modern messaging interface. Built for seamless user interaction.
A comprehensive AI assistant platform allowing users to upload documents, construct custom knowledge domains, and query intelligent chatbots powered by Google Gemini.
Scalable video hosting backend built with Node.js, Express.js, and MongoDB. Complete media pipeline with auth, video transcoding queue, comments, likes, and user subscriptions.
Event-driven Node.js application that consumes RabbitMQ messages and pushes real-time updates to web clients using Socket.IO. Includes resilient reconnection, health monitoring, and Dockerized setup.
Production-style finance backend with role-based access control, JWT authentication, ledger record verification, dashboard analytics endpoints, and robust validation.
Offline-First personal budgeting app built with Next.js following Local-First principles. Works completely offline with auto-save, sync queue, and transaction categorization.
Robust Python pipeline for extracting structured line items from PDFs and documents. Hybrid text extraction with OCR fallback and borderless table detection.