Backend Engineer - AI

Responsibilities:

• This is not a role of maintaining existing systems. You'll build core infrastructure from scratch.
• Agentic Email Intelligence (Freight Forwarding) - Multi-agent pipeline: fetch, classify, extract, auto-respond - LLM-powered extraction of freight rates and shipment details from unstructured emails - Prompt versioning and LLM observability with Langfuse - IMAP/Outlook/Gmail sync handling quotas, reconnection, multi-account
• Distributed Video Processing Pipeline (Warehouse Ops) - Cloud-to-edge architecture: Azure Functions Service Bus on-premise NVR agents - YOLO-based object detection, FFmpeg video clipping, overlays - Event-driven job orchestration with dead-letter handling, retries - Automated evidence generation and email dispatch
• Scalable Backend Systems - Multi-tenant FastAPI services serving enterprise clients - PostgreSQL connection pooling, query optimisation, async workers - Azure Container Apps, Functions, Blob Storage - CI/CD with GitHub Actions, same-day production deploys
• You'll own features end-to-end from database schema to API to deployment.

Requirements:

• Experience with large-scale async systems (queues, workers, and event-driven architecture).
• Track record of productionizing LLM applications (not just Langchain tutorials).
• Comfort with AI coding tools (Cursor, Claude Code): We use these daily.
• With systems thinking, you design for scale, failure modes, and end-to-end product impact.
• You say I built it; no, my team built it. We want drivers, not passengers.
• Thrived in early-stage startups where ambiguity is the norm and speed matters.
• 2-4 years of production backend experience. You've built and maintained systems that real users depend on. Not hobby projects.
• Strong system design fundamentals. You can design a multi-tenant SaaS backend, think about connection pooling, handle async job queues, and debug production issues at 11 PM.
• Deep Python knowledge Not just that I can write Python. We want someone who understands generators, context managers, async/await internals, metaclasses, and the stuff that separates good code from great code.
• Database optimisation experience: You've dealt with slow queries, missing indexes, connection pool exhaustion, and N+1 problems. You know how to read, EXPLAIN, and ANALYZE.
• Cloud deployment experience: You've deployed services on AWS/Azure/GCP. You understand when to use serverless vs. containers and how to set up queues, manage secrets, and configure autoscaling.
• LLMOps exposure: You've taken an LLM-based feature from prototype to production. You understand prompt versioning, evals for accuracy, output validation, cost optimisation, and handling hallucinations.

Tech Stack:

• AI/LLM: Claude, OpenAI, Langfuse (LLM observability), and agentic workflows.
• Computer Vision: YOLOv8 OpenCV, FFmpeg.
• Backend: Python 3.11+, FastAPI, async PostgreSQL, SQLAlchemy.
• Cloud: Azure (Container Apps, Functions, Service Bus, Blob Storage).
• Frontend: React 18 TypeScript, Vite.
• DevOps: Docker, GitHub Actions, and Application Insights.