6+ years shipping production GenAI and high-throughput backends in Python (FastAPI, Django) · Bengaluru, India 🇮🇳
🟢 Open to: Senior Backend / AI Engineer roles (remote or Bengaluru) · Freelance & consulting projects
I build AI products that survive production — not demos.
- 🤖 LLM applications — RAG pipelines, agentic workflows & tool calling, structured outputs, guardrails
- 🔎 Search — hybrid retrieval (BM25 + vector kNN + RRF), re-ranking, semantic search on Elasticsearch / Pinecone / pgvector
- 📉 LLM cost & latency — semantic caching, batching, model routing, multi-provider fallback (OpenAI, Anthropic, open-source)
- 🧪 LLMOps & evaluation — golden-set regression evals, LLM-as-judge in CI, token/latency/cost tracing
- ⚙️ Distributed backends — Kafka, RabbitMQ, Celery, Redis, event sourcing, idempotency, circuit breakers
- 🏗️ End-to-end ownership — architecture → build → deploy (Docker, Kubernetes, AWS/Azure) → on-call
| Project | What I built | Result |
|---|---|---|
| LLM Review Intelligence (US client, 75K+ properties) | Async LLM classification pipeline — topic, sentiment, urgency | 500K+ reviews/month, ~60% lower LLM cost, p95 < 2s |
| "Ask your portfolio" RAG assistant | Hybrid retrieval + re-ranking with cited answers | Survey free-text auto-summarised into themes & actions |
| AI review-response drafting | RAG + brand voice + PII / fair-housing guardrails, human-in-the-loop | Drafting time 10 min → < 1 min |
| Magic Quote (B2B construction commerce) | OCR → LLM extraction → hybrid Elasticsearch matching on 10K+ SKUs | Quote turnaround hours → minutes |
| mobAI commerce assistant | RAG on Pinecone + OpenAI | –30% support tickets, +18% conversion |
| Semantic product search | Embeddings + Elasticsearch, Redis caching at 10K+ QPS | +40% match accuracy, –35% zero-result searches, –35% latency |
| Event-driven backbone | Kafka across 8 microservices, 50K+ events/day, DLQs | 99.9% uptime, p95 < 100 ms |
| Fintech: BNPL & loyalty | Interest/penalty engine, Razorpay, event-sourced loyalty ledger | Live credit + rewards for B2B buyers |
GenAI & LLM: OpenAI · Anthropic Claude · LLaMA · Groq · LangChain · RAG · Agents · Prompt Engineering · Embeddings · Re-ranking · OCR Search & Vector DB: Elasticsearch · Pinecone · pgvector Backend: Python · FastAPI · Django · DRF · Node.js · TypeScript · REST · Microservices · WebSockets Messaging & Async: Kafka · RabbitMQ · Celery · Redis Data & Cloud: PostgreSQL · MongoDB · Docker · Kubernetes · AWS · Azure · CI/CD · Observability
- Senior Backend & AI Engineer (Feb 2025 – Present) · Madoverbuildings — LLM review intelligence, RAG, agentic triage, LLMOps on multi-tenant FastAPI + Kubernetes
- Senior Backend Engineer / SDE II — Supersourcing (Jul 2022 – Feb 2025) · Client: Mad Over Buildings (Hindware-backed B2B construction commerce) — Magic Quote, mobAI, semantic search, Kafka backbone, BNPL & loyalty, checkout/RFQ/orders at 5K+ orders/day
- Software Engineer — Supersourcing (May 2020 – Jun 2022) · Client: INQ (omni-channel telecom) — billing APIs for 100K+ invoices/month, enterprise SSO with Keycloak (SAML/OAuth2)
🎓 MCA — Madan Mohan Malaviya University of Technology, Gorakhpur
If you need any of these, let's talk:
- ✅ RAG chatbot / AI assistant over your documents, products or support data
- ✅ AI document processing — OCR + LLM extraction (quotes, invoices, forms → structured data)
- ✅ Cutting your LLM bill — caching, routing, batching, evals
- ✅ Semantic / hybrid search for e-commerce catalogues
- ✅ Scaling a Python backend — FastAPI/Django, Kafka, Celery, Redis, Kubernetes
📩 abhijeetg40@gmail.com · LinkedIn
100K+ on YouTube · 120K+ on Instagram — I make advanced backend topics simple. Popular: Django + Kafka & Zookeeper · Celery Beat schedulers · Real-time location tracking with Google Maps & Kafka
👉 YouTube · Instagram @_abhijeetgupta