I build autonomous systems that turn hours of manual work into seconds.
AI Engineer specializing in multi-agent orchestration, GraphRAG, LLM evaluations, and enterprise workflow automation — engineering deterministic, production-grade pipelines with Python, LangGraph, MCP, and Next.js.
3rd Year
B.Tech IT student
4+
hackathon finishes
Kauvery Hospitals
AI & Automation Intern
A simplified live view of how agents pass context, route tool calls via MCP, and checkpoint state autonomously — the same pattern used across the pipelines below.
Where manual work goes to end.
Three core capabilities, applied together, to replace repetitive human effort with autonomous, auditable systems.
Multi-Agent Orchestration
Designing scalable, deterministic agent graphs with LangGraph and CrewAI — where specialized sub-agents collaborate with state memory, dynamic routing, and human-in-the-loop controls.
GraphRAG & Hybrid Retrieval
Architecting context-rich retrieval pipelines integrating vector embeddings (FAISS/ChromaDB), knowledge graphs, and hybrid re-ranking to deliver zero-fabrication factual synthesis.
Automated Scraping & Intelligence
Deploying resilient web scraping engines with Playwright and Scrapy alongside n8n automations to extract high-yield competitor data and sync with enterprise APIs.
Production AI Stack & Autonomous Infrastructure
Production tools engineered daily for multi-agent reasoning, resilient data scraping, and high-throughput full-stack systems.
AI & LLM Engineering
Web Scraping & Data Extraction (SEO & Intelligence)
Full Stack & Web Frameworks
Databases, Cloud & DevOps
From hours of manual work to seconds of autonomous processing.
Real pipelines, real time saved. This section updates as new projects ship — built to grow with the work.
BillAgent Pro — Multi-Agent Expense Intelligence
AI bill management system leveraging Gemini 3.7 Flash Vision. Features a multi-agent pipeline (OCR -> Validation -> Learning) to extract vendor, line items, totals, and tax with confidence scoring.
Manual
data entry & line extraction
~2-3 sec
instant AI OCR & validation
Multi-Agent
Validation & Learning loops
JWT Auth
Isolated user tenant data
MediCode AI — EHR & Medical Coding Platform
AI medical coding platform automating ICD-11, CPT, and SNOMED CT code generation from clinical notes. Validates HL7 FHIR R4 compliance, generates SOAP summaries.
Manual
clinical coding & audit
Instant
ICD-11 & CPT extraction
HL7 FHIR
R4 compliant validation
Dockerized
Full-stack container setup
PolicyPulse AI — Autonomous Compliance Platform
Autonomous policy compliance platform. Ingests GDPR/SOC 2 policy PDFs, scans databases, and runs a ReAct Agent loop alongside 4 specialist agents.
Months
undetected policy breaches
Autonomous
real-time scanning & fix
GDG Delhi
HACKFEST Top Project Winner
ReAct Loop
4 Specialist AI Agents
Aura-NPU — Offline Neuro-Inclusive Assistant
Multimodal AI assistant running 100% offline on the AMD Ryzen AI 300 Series XDNA 2 NPU (~100ms latency). Supports 22 Indian languages.
Cloud Only
English-only AI dependency
~100 ms
100% local NPU edge inference
22 Languages
Full Indian language set
Zero Cloud
100% offline privacy safe
Medi-Flow India — National Blood Supply Network
Client-side national blood supply coordination network across 12 Indian cities. Uses a 4-step AI Agent pipeline to monitor shortages and route ambulances.
Critical
uncoordinated blood shortages
Real-time
AI stock watch & smart routing
46 Hospitals
Across 12 major Indian cities
4 AI Agents
Autonomous supply routing
MarketMind AI — Multi-Agent Market Intelligence
Multi-agent platform for small businesses. Runs a 4-agent LangGraph StateGraph pipeline to scrape competitor sites, perform SWOT, and dispatch WhatsApp alerts.
Manual
competitor & market tracking
1-Click
full strategy & 24hr social plan
LangGraph
4-agent sequential workflow
Automated alert dispatches
Simulate a LangGraph Multi-Agent Workflow
Select a query below to inspect an autonomous agent loop in action — routing intents, executing MCP retrieval tools, applying Pydantic guardrails, and calculating real-time LLM evaluation scores.
Router Node
MCP Tool Call
Verification Agent
Synthesizer Node
A small window into how these agents run.
Trigger a simulated pipeline below. Nothing here calls a real backend — it's a preview of the interaction pattern used in production.
Select a demo below to trigger a simulated run →
Still 3rd year. Already shipping to production.
I'm currently a 3rd-year B.Tech Information Technology student at SNS College of Technology (CGPA 8.47), working as an AI & Automation Engineering Intern at Kauvery Hospitals — building AI-driven automation pipelines, competitive intelligence systems, and digital strategy tooling for hospital leadership.
My focus is using LLMs for deterministic reasoning and orchestration rather than content generation: multi-agent systems with LangGraph, GraphRAG retrieval with FAISS/ChromaDB, Model Context Protocol (MCP) integrations, and LLM evaluations with RAGAS and LangSmith.
I also engineer robust web scraping and headless browser pipelines (Playwright, Scrapy, BeautifulSoup) to extract high-signal structured intelligence from across the open web.
Outside of engineering, I care about photography, video editing, and visual design — which translates into how I build software: deliberate, mathematically grounded, and free of clutter.