AI Engineer (Backend)
Role Overview:
We're seeking an experienced AI Engineer with strong backend engineering skills to build and operate the AI agents and Python services behind our property-management platform. This role is responsible for developing agent-based workflows, the FastAPI services around them, and the evaluation, guardrails and cost controls that make AI output reliable enough for production use in an audit-sensitive domain.
Key Responsibilities:
Design, build and extend agent workflows — supervisor routing, tool calling, human-in-the-loop approval, guardrails and fallback chains.
Develop and maintain FastAPI backend services with REST and SSE streaming, async SQLAlchemy, Alembic migrations and JWT-based authentication.
Implement RAG retrieval and grounding over technical documentation and historical work orders, so AI recommendations can cite their evidence.
Build and own the evaluation harness — replay testing against golden traces, fake-LLM unit tests, and behaviour-drift detection in CI.
Manage AI cost and reliability controls — per-tenant token budgets, rate limits, usage tracking and multi-model routing through the model gateway.
Define and version the schemas and API contracts exchanged between agents and services, with automated contract testing against OpenAPI specs.
Integrate AI agents with ERP and third-party systems, and contribute to the spec-driven code generation pipeline.
Required Skills & Experience:
Strong Python engineering with production services — async, typing, testing and packaging.
Hands-on production experience with LLM orchestration frameworks (LangGraph, LangChain or equivalent), tool calling and structured output.
Proven experience evaluating non-deterministic systems — eval harnesses, replay testing and regression detection.
Solid backend and API design skills — schema design, versioning, REST/SSE and event-driven patterns.
Strong relational database skills (PostgreSQL preferred) beyond ORM basics, including query performance analysis.
Working knowledge of AWS (ECS Fargate, Bedrock, S3, SQS/SNS) and CI/CD practices.
5–8 yearsin software engineering, including at least 18 months building production LLM/AI systems.