AI Software Engineer / Bengaluru

Vansh
Bhatnagar

I build production agentic AI systems — LangGraph orchestration, MCP tool calling and RAG pipelines that survive real traffic.

Open to relocation & remote
ask-vansh

Ask about my work — the agentic system I shipped at 9AI, how I build RAG pipelines, or why you should hire me. It retrieves over my own writing, so it only knows what I've actually done.

Experience

Where I've worked

Jun 2026 — Present

Current

Junior Software Developer

NinjaHire

  • Ship platform features end-to-end — the backend service and the workflow built on top of it — and own them past the merge: design, review, deploy, then watch them run.
  • Work client-reported production defects, chasing the ones that keep recurring instead of patching each symptom as it lands.
  • Turn vague pain points from product and support into scoped work that actually ships.

Sep 2025 — Jun 2026

SDE Intern — AI

9AI

lower response latency
40%
downstream systems served
3+
less manual review
30%
  • Architected and shipped an end-to-end agentic AI system using LangGraph orchestration, Model Context Protocol tool calling and RAG pipelines backed by Pinecone and Chroma — automating sales lead qualification and cutting response latency by 40%.
  • Built and deployed AI/LLM backend services as REST APIs in Python (FastAPI), containerized with Docker and released through CI/CD, serving 3+ downstream business systems and enabling real-time lead scoring.
  • Applied chain-of-thought, few-shot and role-based prompt engineering across multi-agent workflows, reducing manual review overhead by ~30% while improving output consistency at scale.
  • Integrated commercial LLM provider APIs into existing business systems, applying retry, rate-limiting and observability patterns to keep AI workflows reliable in production.

2025

AI/ML Intern

NJR I3 Labs

more engagement
35%
better retention
20%
recommendation accuracy
85%
  • Developed an interactive 3D learning interface with Three.js that increased user engagement by 35% and improved knowledge retention by 20% through immersive visualization of complex concepts.
  • Implemented AI-driven content personalization with Gemini, reaching 85% recommendation accuracy.
  • Cut average learning completion time by 30% through intelligent learning-path optimization.

Selected work

What I've built

LangGraph Multi-Agent Cybersecurity Platform

Multi-agent orchestration

2026

A multi-agent system that runs offensive-security tooling in parallel, grounding each agent in retrieved context so scans stay targeted instead of exhaustive.

1,500+ findings

SaleSpeak

RAG-powered conversation

2025

A conversational agent that helps people make better purchasing decisions through natural voice interaction and retrieval-grounded product recommendations.

AI-Tutor

Retrieval + curriculum

2025

An educational platform that scrapes learning material from across the web, processes it with Gemini, and assembles personalized learning paths.

Grass & Milk Production Predictor

Applied computer vision

2024

A computer-vision system that reads farm photographs to grade pasture quality and forecast downstream milk yield.

Capabilities

What I work with

Agentic AI & LLM

RAG, Vector DBs & Providers

Languages & Backend

Cloud, DevOps & Distributed Systems

Data & ML

Education

Where I studied

Sep 2022 — May 2026

B.Tech, Computer Science

Techno India NJR Institute of Technology · Udaipur, India

Relevant coursework: Machine Learning, Distributed Systems, Software Engineering

CGPA 8.65 / 10

Recognition

Awards & honours

International2024

WHO Letter of Recognition

Recognized by the World Health Organization for building a real-time, geo-tagged monitoring platform for tracking unvaccinated children.

NIT Goa2024

1st Place — OxGenIgnite Hackathon

Led the team that built a fiat-to-fiat crypto swap on Stellar using Soroban smart contracts and the Stellar DEX, cutting cross-border transaction cost 5–15% through full fee transparency.

Hackathon2024

2nd Place — CodeRed 4.0 Hackathon

Led the team that built an AI diagnostic system for rural healthcare, training a deep-learning CNN on a 10,000+ image X-ray and CT dataset to identify medical conditions.

Contact

Let's talk

I'm open to roles building production agentic AI — and to conversations about LLM systems generally, whether or not there's a job attached.

vanshbhatnagar445@gmail.com+91 97853 66298

Bengaluru, India · Open to relocation & remote