Harsh Seth

At Tata Consultancy Services, I build LangChain-based RAG and backend systems for enterprise document workflows. I engineered a document Q&A pipeline that reduced manual document-processing time by 90% and built RESTful APIs connecting RPA automation with LLM components.

I also architected a multi-turn conversational assistant with persistent context and semantic retrieval over ChromaDB, delivering real-time answers from internal knowledge bases. My work includes UiPath RPA automation that improves reliability and reduces operational overhead across business units.

In DocuMind, I built an agentic RAG document assistant using LangGraph, LangChain, FastAPI, ChromaDB, PostgreSQL, SQLAlchemy, and Groq. The agent routes autonomously between search, summarization, comparison, and document-listing tools, while hybrid retrieval, reranking, and confidence thresholds help prevent hallucinated answers.

I'm continuing my MCA while building stateful ReAct agents, persistent-memory systems, and multi-agent AI projects. I’ve also solved 600+ DSA problems, with daily C++ practice focused on graphs, trees, and dynamic programming.