I am streamlining our healthcare claims approval flow so that every repetitive step is handled by code, not people. The goal is simple: remove as much manual work as possible while keeping decisions fast, traceable, and accurate. Here is where the project stands and what I still need: • Core stack. Django REST Framework runs the API layer. Celery workers, brokered by RabbitMQ, will handle multiprocessing and threading so the system scales horizontally. • Generative AI. Azure-hosted OpenAI models deliver the decision logic; traffic must respect an existing token-rate-limit load-balancer. • Data intake. Claims arrive as unstructured documents—PDFs, emails, images. They must be classified, parsed, and written to PostgreSQL. Proper indexing is required for both speed and future analytics. • Data visualisation. A lightweight module should surface real-time KPIs on approval rates and processing times, giving stakeholders instant insight into system health. You will design the Celery task graph, code the GenAI interaction layer, tune PostgreSQL, and expose everything through clean REST endpoints. Customer-inquiry or policy-creation flows aren’t in scope right now, but the architecture should leave room for those extensions later. Acceptance criteria • Docker-compose spin-up of the full stack • End-to-end test: upload claim → automatic approve/reject response • Sub-60-second average processing for a five-page PDF • 95 % unit-test coverage on core business logic If you have deep experience with Django, Celery, Azure, PostgreSQL, and prompt engineering, this should feel right in your wheelhouse.