Shipped Systems

An AI learning assistant that turns YouTube videos, documents, and study materials into instant summaries, smart quizzes, and clear cited answers - so students understand and revise fast, without rewatching lectures or rereading pages. Under the hood it is a production RAG system: hybrid search on Pinecone, multi-LLM routing, and real-time SSE streaming across a large content corpus.

Serves streaming answers from a 20K+ content corpus with sub-2-second first-token latency.

An open-source AI tool that lets anyone query a database in plain English instead of SQL. Ops, finance, and HR teams can self-serve the numbers they need just by asking - no analyst bottleneck, no code - and it draws the right chart for each answer. Built as a LangGraph + OpenAI agent that converts natural language to SQL, holds multi-turn context, remembers with Mem0, and runs in three deployment modes.

Open-sourced under AGPL v3. Architecture scales to enterprise-grade query workloads.

Engineered enterprise-grade Retrieval Augmented Generation pipelines with hybrid search, optimized embedding pipelines, context-aware retrieval, and metadata-driven search architectures for international brand workloads.

Designed and built an AI agent capable of searching across multiple Pinecone indexes simultaneously. Created a universal metadata parser to normalize product fields across diverse catalogs - the same pattern enterprises need when RAG has to span business units with different schemas.

Built production AI assistants for multiple international brands - product recommendations, multilingual support, appointment booking, complaint handling, and FAQ automation. Fine-tuned models using custom JSONL datasets with ReAct reasoning chains.

A trading platform that helps traders test ideas and break bad habits. Backtest any strategy on four years of real, broker-exact data, and every trade is auto-journaled with the emotion behind it - so the platform can flag patterns like revenge trading or overtrading and help cut them out. One-click execution runs through the trader's own broker (cTrader, IC Markets, Fyers), with a premium Flow tier for order-flow visualization and AI-assisted algo building. Built full-stack with MetaAPI + Fyers integration, WebSocket streaming, a 3-worker Celery architecture, and AWS.

A trading-education program that teaches rule-based ("mechanical") trading - following predetermined rules instead of trading on emotion. Students learn six proprietary systems for forex, crypto, and Indian markets through self-paced courses, monthly 5-day live execution labs, a paper-trading arena, and a performance-based Certified Mechanical Trader credential. Built full-stack with live streaming, authentication, and payments on Firebase.