How akshaypkulkarni/doc-intelligence-platform works
Project type and purpose
• This is a cloud-native, full-stack document intelligence platform with a RESTful API (no separate UI code) that processes, analyzes, and semantically searches documents using AI.
• It exposes FastAPI endpoints, runs background tasks with Celery, persists in PostgreSQL, caches/queues in Redis, stores objects in S3, and integrates with OpenAI and Pinecone.
File structure
• Top-level:
• docker/, docker-compose.yml → containerization
• src/ai_document_platform/ → main application
• scripts/init-db.sql → DB initialization
• tests/ → unit tests
• pyproject.toml, .pre-commit-config.yaml → Python tooling/config
• Inside src/ai_document_platform:
• api/ (FastAPI routes)
• services/ (business logic)
• repositories/ (DB ORM via SQLAlchemy)
• clients/ (external services: OpenAI, Pinecone, S3)
• models/ (Pydantic schemas, data models)
• utils/ (helpers)
README for additional insights
• Technology stack calls out: FastAPI, Celery, PostgreSQL, Redis, AWS S3, OpenAI API, Pinecone, Docker/Kubernetes.
• Architecture is described as microservice-like with API + background workers + external integrations.
Elements
• Clear component labels (e.g., “FastAPI API”, “Celery Worker”)
• Client → API → DB
• API → Redis → Worker → External Services