Nirajan K.

Python Backend Developer | Generative AI | LLM | AI Agents | Claude

I build RAG pipelines and LLM systems that make it to production — not demos that stall. AI that retrieves accurately from your own data, and integrations built to reduce cost, not inflate it. Top Rated | 100% Job Success | 104+ projects delivered. I'm a Python and AI engineer with 7 years building production systems end to end: RAG pipelines grounded in internal documents, AI agents that automate real workflows, trading platforms with real-time data feeds, and full-stack applications taken from empty repo to deployment. SELECTED RESULTS - Cut GPT inference cost 41% via a custom RAG architecture eliminating redundant API calls — no accuracy loss - Reduced API latency 72% through an async redesign on FastAPI + Redis - Sustained 12,000 requests/min at 99.99% uptime after decomposing a Django monolith into microservices — zero-downtime migration - Processed 2M+ events/day at 37ms average latency on containerized infrastructure with PostgreSQL sharding - Compressed release cycles from 2 days to 2 hours with GitHub Actions CI/CD - Cut AWS spend 35% through infrastructure rightsizing WHAT I BUILD RAG systems that retrieve accurately from your data. AI agents automating real workflows. Cost-aware LLM integrations. Plus the full stack around them: Python backends, React/Next.js frontends, REST APIs in Django or FastAPI, real-time WebSocket features, cloud deployment on AWS/Azure/GCP, and CI/CD pipelines. THE STACK AI/LLM: GPT, Claude, Gemini, LangChain, LlamaIndex, RAG, AI Agents, pgvector, Pinecone, Qdrant Backend: Django, DRF, FastAPI, Flask, Celery, AsyncIO, WebSockets Frontend: React, Next.js (App Router), TypeScript, Tailwind CSS Databases: PostgreSQL, MySQL, MongoDB, Redis, Elasticsearch, Supabase DevOps: AWS, GCP, Azure, Docker, Kubernetes, Nginx, GitHub Actions Linux: Ubuntu, Debian, VPS hardening, SSL, Systemd If you're shipping an AI feature or need a backend that holds up in production, send a message or an invite. I reply within a few hours.