Available for backend engineering roles

Shubham
Bhattacharya

$ Backend & AI Engineer

I build autonomous AI agents and production backend systems that handle real money, real users, and real failure modes. I treat every AI output as untrusted input.

View My Work Let’s Talk
01

About

Shubham Bhattacharya

I'm Shubham — a backend engineer from Vadodara, India, obsessed with building systems that survive contact with real users, real money, and real failure modes.

I graduated with a BCA from Parul University and spent the last two years going deep on agentic AI systems, production APIs, and backend architecture. Not through courses — through shipping.

My engineering philosophy: "Treat AI as an untrusted API." Every AI response passes through strict Zod schema validation before touching a database.

When I'm not building systems, I'm documenting my learning journey publicly — 60+ days of hands-on work covering RAG, vector databases, agent orchestration, and production LLMOps.

02

Projects

Production-grade systems I've built, not tutorial follow-alongs.

Gigi AI Ops

Agentic AI · Backend

Production-grade agentic AI backend for Gigi Energy automating e-commerce ops — cancellations, refunds, Q&A — with zero autonomous financial risk.

  • Decision EngineLangGraph.js state machine routes conversations dynamically
  • Safety GateZod runtime validation on all LLM outputs — crash rate reduced to 0%
  • Cost Trackertiktoken WASM calculator — USD/INR per-request tracking
  • Financial GuardHITL threshold: refunds >₹200 auto-held for admin validation
LangGraph.jsZodpgvectorPostgreSQLBullMQRedis
Source Code

Medo Copilot

AI · Full Stack

AI-powered medical assistant with document analysis, patient query handling, and clinical decision support using RAG.

  • RAG PipelineDocument ingestion → embedding → pgvector retrieval → structured output
  • SecurityContext injection hardened against prompt spoofing
  • RealtimeServer-Sent Events for streaming responses
Node.jsTypeScriptNext.jspgvectorSSEPrisma
Source Code

Session-Secure Context API

Security · Backend

Reusable middleware for AI agent APIs preventing prompt injection and context spoofing. HMAC-signed session tokens with Zod validation.

  • SecurityHMAC-signed tokens — context can't be forged client-side
  • MiddlewareDrop-in Express.js middleware — 3 lines to secure any endpoint
Node.jsTypeScriptHMACZodExpress.js
Source Code

Multi-Tenant Job Queue

Infrastructure · Backend

Multi-tenant background job processing on BullMQ and Redis with tenant isolation, retry logic, and graceful shutdown.

  • IsolationPer-tenant Redis queues with namespace prefixing
  • ReliabilityExponential backoff, dead letter queues, graceful SIGTERM
  • ObservabilitySentry + Pino structured logging on every worker event
BullMQRedisNode.jsTypeScriptSentryDocker
Source Code
03

Stack

Technologies I use to ship production systems.

Runtime & Language

Node.jsTypeScriptJavaScriptPython

AI & Agents

LangGraph.jspgvectortiktokenOpenAI SDK

Database & Cache

PostgreSQLRedisPrismaMongoDB

Backend & APIs

Express.jsREST APIsBullMQSocket.io

Frontend

Next.jsReactTailwind CSSSSE

Security & Validation

ZodClerkJWTHelmet

DevOps & Infra

DockerGitHub ActionsAWSVercel

Observability

SentryPinoPostman
04

Journey

Agentic AI & Backend Engineering

2024 — Present

Independent Study & Open Source

  • Built multi-step AI agent pipelines with Groq Llama models, pgvector retrieval, and BullMQ job queues
  • Implemented Zod runtime validation on all LLM outputs to prevent untrusted model responses from crashing downstream
  • Engineered session-secure context injection to prevent prompt spoofing in production AI agents
  • Practiced production backend patterns: graceful shutdowns, multi-tenant DB schemas, rate limiting

60-Day Agentic AI Learning Sprint

2024

Documented Public Learning Journey

  • Documented 60+ days of hands-on work covering RAG, embeddings, vector databases, agent orchestration
  • Built and shipped real projects each week — not tutorials, but original systems solving specific problems

BCA — Parul University

2021 — 2024

Bachelor of Computer Applications

05

Get In Touch

I build production AI pipelines and backend systems. Currently looking for backend engineering roles — internships, junior positions, or remote startup work where I can ship production code from day one.

Email Me Download Resume
Emailshubhambhattacharya107@gmail.com GitHubgithub.com/shubhambhattacharya-dev LinkedInlinkedin.com/in/shubhambhattadev X@Shubham_level
Email Me Resume