I build production-ready AI systems from stateful multi-agent orchestration to advanced RAG pipelines. I specialize in turning complex AI research into real, shippable products.

• Open-source contributor at CNCF (LFX Mentee) and Google Summer of Code (KDE).
• Competitive Programmer Codeforces Specialist (1481), ICPC India Rank 1380.
• Winner of ISRO Space Hackathon | Runner Up at Ethos Hackathon (IIT Guwahati) 2024.
• BTech CSE @ IIIT Agartala (CGPA: 8.13) Graduating 2026.

AI / LLM

Languages & Databases

Frameworks & Infrastructure

A multi-layered RAG system for safe, authentic Ayurveda content generation. Built a 6-agent orchestration framework using LangGraph to automate domain-specific content creation while enforcing strict medical authenticity.

• Hybrid Search Architecture: Implemented BM25 and Vector search (ChromaDB) for high-precision retrieval.
• CRAG (Corrective RAG): Integrated a self-healing loop that evaluates retrieval quality and re-queries for missing context when needed.
• Knowledge Graph Constraints: Utilized Graph-based grounding to prevent hallucinations and ensure clinical alignment.
• Python · LangGraph · ChromaDB · FastAPI · CRAG · Knowledge Graphs · Gemini

Voice-based mock interview platform with grounded multi-agent evaluation and per-call LLM observability

• Hybrid RAG-grounded Agents: A LangGraph state machine (interviewer → evaluator → summary) reads from a hybrid pgvector + BM25 retriever over the candidate's resume and a tagged question bank, with Gemini 2.5 Flash structured outputs for typed scoring.
• Real-time Voice Loop: Full-duplex browser WebSocket pipes ElevenLabs Scribe transcripts into the graph and streams Edge TTS audio back as binary frames, with webcam, anti-cheat focus tracking, and an in-page CodeMirror editor for coding turns.
• Production-grade LLM Ops: Every model call flows through one helper emitting OpenTelemetry GenAI spans with token counts and USD cost, retries on quota with API-key rotation, and write-throughs to a Postgres table powering an admin /llm-usage dashboard.
• Next.js 16 · React 19 · TypeScript · FastAPI · LangGraph · Gemini 2.5 · pgvector · PostgreSQL · Redis · ElevenLabs · OpenTelemetry · LangSmith

Visualizing Software Supply Chain Security (CNCF - LFX '25)

• Metadata Auditability: Built an interactive dashboard to audit The Update Framework (TUF) metadata, enhancing transparency in secured supply chains.
• Interactive Visuals: Implemented hierarchical tree views, version diffing, and metadata breakdown using React & D3.js.
• Next.js · React · TypeScript · D3.js · TailwindCSS

Automated JioNet Campus WiFi connectivity.

• Lightweight script to bypass repetitive login portals and ensure persistent internet connectivity in university high-density networks.
• Python · Automation

Mar 2025 – May 2025 · Remote

• Built an interactive TUF Metadata Visualizer with Next.js & React to audit secured software supply chains.
• Implemented tree visualizations, expandable tables, and version diff views for metadata auditability.

Jun 2025 – Jul 2025 · Remote

• Engineered the official KDE Linux (Project Banana) website using Hugo with multilingual support & dark/light modes.

Mar 2024 – Feb 2025 · Remote

• Built a unified production dashboard with React, TypeScript, Tailwind, improving usability.
• Integrated Gong APIs for real-time conference call insights.
• Managed CI/CD and testing across a Django REST + PostgreSQL stack.

• ICPC India Prelims 2023: Global Rank 1380
• Meta Hacker Cup R2: Global Rank 3532
• Google Code Jam: Rank 4339
• CP Titan 3.0, 4.0 & 5.0: 2nd Position