| AI Engineer · Building multi-agent systems & RAG pipelines | Portfolio · LinkedIn · X · GitHub |
I build AI systems that actually ship — multi-agent workflows, self-correcting RAG architectures, and agentic platforms.
- Integrated M.Tech AI @ VIT Bhopal · 9.33 CGPA · specializing in AI & Machine Learning
- Maintainer of awesome-langgraph · curated patterns & resources for LangGraph agentic systems
Currently building GroundedAI, MailMind, and rzp Merchant.
|
GroundedAI (Self-Correcting RAG) |
Multi-source RAG agent using a Corrective RAG (CRAG) loop to detect and correct hallucinations. • pgvector + tsvector hybrid search with Reciprocal Rank Fusion (RRF) • Local NLI evaluation (DeBERTa-v3) for zero-cost faithfulness checks • Stack: LangGraph, FastAPI, pgvector, Next.js |
|
MailMind (Agentic Email Processing) |
Autonomous email processing engine that reads, classifies, and acts on incoming emails. • Event-driven ingestion via Resend webhooks & LLM intent classification • Production Human-in-the-Loop (HITL) approval gate for sensitive actions • Stack: LangGraph, FastAPI, FAISS, PostgreSQL, Next.js 16 |
|
rzp Merchant (Agentic E-Commerce Platform) |
AI-native merchant platform built for the **Razorpay Buildathon**. • 3-Tier Agent Security supporting Model Context Protocol (MCP) & spending mandates • "Shop with AI" conversational agent, WebAuthn passkeys, and Drizzle ORM • Stack: Next.js 16, TypeScript, Gemini 2.5, Drizzle, Better Auth |