I’m leading the digital programme at a German construction company and we’re replacing a tangle of spreadsheets with a unified, API-driven platform for data analysis and reporting. Your main mission is to implement the calculation engines that power our new cost-estimation service and to embed them in production-ready FastAPI micro-services. Tech landscape • Python 3.11, FastAPI, Pydantic v2 • LightGBM, XGBoost and Prophet for the predictive bits • PostgreSQL 15 with pgvector extensions for feature storage • Celery & Redis for asynchronous workloads • Everything ships as Docker containers, built and deployed via GitLab CI What you’ll do – Translate the specification and signed API contracts into clean, tested, well-documented code – Design and implement cost-estimation models, plus supporting data pipelines and calculation logic – Optimise queries and vector operations in PostgreSQL – Package the service, write Helm-friendly Dockerfiles, and hand over a one-command local dev environment Deliverables 1. Source code with ≥90 % unit-test coverage 2. Docker image and compose stack for local run 3. Auto-generated API docs (OpenAPI) 4. Short setup & operations guide (Markdown) Acceptance criteria • All endpoints pass the supplied Postman test suite • Model accuracy meets or exceeds the baselines stated in the spec • CI pipeline builds, tests and pushes the image without manual steps To apply, show me past work that demonstrates the mix of Python, FastAPI, and ML in a production context. A link to a public repo or a concise write-up of a private project is perfect. Domain experience in engineering or manufacturing will definitely help you stand out. Looking forward to seeing how you’d tackle this modernisation effort!