00 — Available for Engineering RolesShimoga, Karnataka / Kanpur, UP

Backend Developer | Generative AI & Agentic AI

Backend Developer & Agentic AI.

Engineering scalable backend architectures, autonomous AI agents with LangGraph & LangChain, and production-ready RAG workflows.

Résumé ↓

01 — About

Engineering scalable systems & agents.

Bio & Summary

Hi, I'm Shivam.

Backend Developer specializing in Generative AI & Agentic AI. Engineering scalable backend architectures, autonomous AI agents with LangGraph & LangChain, and production RAG systems.

Experienced in designing end-to-end full-stack architectures, Joi-validated REST APIs, and containerized microservices.

Résumé ↓GitHub ↗LinkedIn ↗

Shivam Chaudhary · AI & Machine Learning · Shivam Chaudhary ·

DEVELOPER ID

Shivam Chaudhary

Backend Developer | Generative AI & Agentic AI

ID NO.SC-AIML-2024

DEPT.AI & Machine Learning

VALID TILL2028

WHAT I AM

ROLEBackend Developer · Generative AI & Agentic AI

EDUCATIONB.E. AI & ML — JNNCE (2024–Present)

PRIMARY STACKLangGraph, LangChain, Hybrid RAG, Node.js, FastAPI

KEY WORKSGit RAG Assistant, AI Commander VS Code Agent, EventHub

HONOURS1st Place, TCS Hackathon · 2nd Place, Mysterio 6.0

Shivam Chaudhary

If found, say hello · Shivamchy076@gmail.com

Hover / Tap to flip ID

Quick Facts

Snapshot & Ethos

LocationShimoga, Karnataka / Kanpur, UP

Degree & CollegeB.E. AI & ML · JNNCE

Current ExperienceFreelance Web Developer · Indalnova

EmailShivamchy076@gmail.com

“Architecting scalable backend logic and autonomous AI agents that transform Generative AI into real, reliable production workflows.”

02 — Stack & Skills

The periodic table of my stack.

03 — Selected Work

Things I've built with Agentic AI.

01/Agentic AI Coding Assistant·08/2026 – Present

Git RAG — Agentic GitHub Repository Assistant

Built an Agentic AI GitHub Repository Assistant using LangGraph, Hybrid RAG (Qdrant + BM25), Cross-Encoder reranking, and GitHub MCP as an end-to-end, developer-focused AI coding assistant.

•Implemented JEV-based intent classification and intelligent MCP tool selection, enabling the agent to dynamically decide between normal conversation, repository retrieval, and live GitHub actions.

•Added evidence filtering, deduplication, and local Hugging Face embeddings with grounded LLM generation (Groq/OpenRouter/Google AI).

•High-performance backend served through FastAPI and fully containerized with Docker for seamless deployment.

•Hybrid retrieval combines BM25 keyword matching and Qdrant vector search with Cross-Encoder reranking for code discovery.

Python

LangGraph

LangChain

FastAPI

Docker

Qdrant

Hugging Face

Groq

View on GitHub ↗

git-rag-agent // LangGraph

Illustrative UI

> git_rag.query(“How does hybrid retrieval work in this repo?”)

✓ [JEV Intent Classifier]: RETRIEVAL_AND_TOOL_SELECTION

• Step 1: BM25 Lexical Scan (top 15 chunks)

• Step 2: Qdrant Vector Semantic Match (cosine dist < 0.18)

• Step 3: Cross-Encoder Rerank (score: 0.942)

[Grounded Answer]: Hybrid RAG combines BM25 for precise token matching and Qdrant dense vectors for semantic context, passed through Cross-Encoder reranking before Groq LLM inference.

MCP TOOL: connectedLATENCY: 142ms

Built an Agentic AI GitHub Repository Assistant using LangGraph, Hybrid RAG (Qdrant + BM25), Cross-Encoder reranking, and GitHub MCP as an end-to-end, developer-focused AI coding assistant.

•Implemented JEV-based intent classification and intelligent MCP tool selection, enabling the agent to dynamically decide between normal conversation, repository retrieval, and live GitHub actions.

•Added evidence filtering, deduplication, and local Hugging Face embeddings with grounded LLM generation (Groq/OpenRouter/Google AI).

•High-performance backend served through FastAPI and fully containerized with Docker for seamless deployment.

•Hybrid retrieval combines BM25 keyword matching and Qdrant vector search with Cross-Encoder reranking for code discovery.

git-rag-agent // LangGraph

Illustrative UI

> git_rag.query(“How does hybrid retrieval work in this repo?”)

✓ [JEV Intent Classifier]: RETRIEVAL_AND_TOOL_SELECTION

• Step 1: BM25 Lexical Scan (top 15 chunks)

• Step 2: Qdrant Vector Semantic Match (cosine dist < 0.18)

• Step 3: Cross-Encoder Rerank (score: 0.942)

[Grounded Answer]: Hybrid RAG combines BM25 for precise token matching and Qdrant dense vectors for semantic context, passed through Cross-Encoder reranking before Groq LLM inference.

MCP TOOL: connectedLATENCY: 142ms

PythonLangGraphLangChainFastAPIDockerQdrantHugging FaceGroq

View on GitHub ↗

04 — Experience & Path

Education and experience as one path.

2024Education

09/2024 – Present

Bachelor of Engineering — Artificial Intelligence & Machine Learning

Jawaharlal Nehru New College of Engineering — Shimoga, Karnataka

•Pursuing undergraduate degree in Artificial Intelligence & Machine Learning, focusing on backend engineering, system architecture, distributed services, and autonomous agentic workflows.

•Active developer focusing on Data Structures, Algorithms, and production software architectures.

2025Experience

08/2025 – Present

Freelance Web Developer — Indalnova

Remote / Kanpur, UP

•Designed and built a full-stack skincare e-commerce site end-to-end (HTML, CSS, JavaScript, Node.js, Express.js, MongoDB, Supabase), shipping dynamic product listings and an order-management system as sole developer from requirements through deployment.

•Continuing to maintain a stable, incident-free production environment by resolving 10–15 client-reported bugs and modification requests per quarter, keeping the live storefront (10–15 orders/month) reliable for ongoing customers.

•Own the project fully — client communication, feature scoping, deployment, and monitoring — for this live, client-facing system, active since launch.

HTMLCSSJavaScriptNode.jsExpress.jsMongoDBSupabase

2026Experience

08/2026 – Present

Agentic AI & Systems Engineering

Git RAG & AI Commander Initiatives

•Built Git RAG: an Agentic AI repository assistant with LangGraph, Hybrid RAG (Qdrant + BM25), and GitHub MCP.

•Developed AI Commander: TypeScript VS Code extension capturing terminal error streams for automated debugging.

•Cut database fetch latency by ~85% with Redis query caching for 1,000+ concurrent users on EventHub.

PythonFastAPILangGraphTypeScriptDockerRedis

NEXT STEPImmediate / Forward

Next — Your Team?

Worldwide (Remote / Onsite)

•Open to backend engineering and Generative & Agentic AI roles.

•Ready to architect resilient production APIs, autonomous agent workflows, and scalable data layers.

Let's build together ↗

05 — Milestones & Ranks

Recognition & milestones.

Quantifiable benchmarks from cybersecurity competitions, national hackathons, and algorithmic problem-solving.

4 Benchmarks · Hackathons & Accolades

1st Prize 🏆01 / 04

TCS Hackathon @ JNNCE•TCS Tech Day

1st Place Winner

Secured 1st Place in the prestigious TCS Hackathon hosted at JNNCE as part of TCS Tech Day, solving real-world challenges through agile teamwork and robust engineering.

KEY METRICPlace Award

0st

02 / 04

Bug Bounty Event•Mysterio 6.0

2nd Place Winner

Secured 2nd Place identifying and exploiting live security vulnerabilities in production web applications.

KEY METRICPlace Award

0nd

03 / 04

Competitive Engineering•Hackathons & Contests

6 Hackathons Competed

Participated in 6 hackathons including TCS Hackathon (1st Place), Smart India Hackathon (SIH), IGNITRON 2K25 (GM University), and Alva's Hackathon.

KEY METRICHackathons

Profile ↗04 / 04

Algorithmic Problem Solving•LeetCode

Active Practice

Actively solving Data Structures and Algorithms problems on LeetCode with continuous daily practice.

KEY METRICProblems Solved

CONTINUOUS PROGRESSand counting →

Actively engineering new RAG pipelines, autonomous agents, and algorithmic problem-solving daily.

06 — Contact

Let's buildsomething together.

↗

DIRECT INBOX

Shivamchy076@gmail.com

PHONE (CALL / WHATSAPP)+91 7667736039

GitHub ↗LinkedIn ↗LeetCode ↗Download Résumé ↓

Based in Shimoga, Karnataka / Kanpur, UP