akshit_29
Remote: Yes, remote only. 4-6h daily overlap with US Eastern, full overlap with EU. Contractor or EOR.
Willing to relocate: No
Technologies: Python, Django/DRF, FastAPI, Celery, Kafka, PostgreSQL, Redis, Next.js, LangGraph, OpenAI Agents SDK, MCP, pgvector, Pipecat, AWS, Azure, Kubernetes
Resume/CV: https://akshit.wtf
Email: [email protected]
Backend engineer, around 4 years, now building production AI systems. I learned the boring parts first.
At Spinny (Indian used-car unicorn): scaled a connected-cars backend to 34M+ requests/day, built the dealer-financing pipeline that disburses $3.4M+ a month, and replaced a 20-30 person auction-scheduling team with an event-driven system on Celery and Kafka. Idempotent webhooks with retries cut external API failures by 90%. Nothing glamorous. It just didn't go down.
Last 18 months, mostly contracting:
- A unicorn client had 20-25 people reading vehicle service reports off a portal with no API. Now it's 3-4. LLM extraction plus computer use, low-confidence results routed to a human queue. Saves them around $5,000/month.
- Hindi-first voice agents on Pipecat + Twilio, around 600ms turn latency. Voicemail detection, IVR navigation, the parts demos skip.
- LLM call scoring against a rubric, text-to-SQL, pgvector RAG. Benchmarked four STT providers on real calls before picking one.
- My own product, PeerSeek: 18 Django apps, one shared payments engine, a multi-agent content pipeline where evals send weak drafts back.
Most "AI engineering" is still engineering. Retries, idempotency, a human gate where the model isn't sure. The model is the easy part.
Looking for: a remote backend/AI role at a small team, contract or full-time, where I own a system end to end. Happy to do a paid trial week first.