AI Document Processing Platform

A production-style asynchronous document-processing service built with FastAPI. Documents are uploaded through the API, persisted, queued for background processing, passed through an extractor abstraction, and returned as structured results.

Architecture

Client
  |
  v
FastAPI API ------> PostgreSQL
  |                     ^
  |                     |
  `----> Redis/Celery --> Worker
                           |
                           v
                    Extractor Provider
                    (demo / LLM adapter)

What it demonstrates

• FastAPI API design
• asynchronous document workflows
• background workers with Celery
• PostgreSQL persistence
• file-storage abstraction
• provider-based extraction architecture
• structured extraction results
• Docker Compose development environment
• automated tests and CI

The default extractor is deterministic so the repository runs without external API credentials. A production LLM or document-intelligence provider can be implemented behind the same interface.

Tech stack

Python FastAPI SQLAlchemy PostgreSQL Redis Celery Docker Pytest

Quick start

cp .env.example .env
docker compose up --build

API docs: http://localhost:8000/docs

Typical workflow

1. Upload a document.
2. The API creates a processing record.
3. A background job processes the document.
4. The configured extractor returns structured data.
5. Status and results are persisted and available through the API.

Engineering focus

This repository focuses on the architecture around AI/document processing rather than hiding everything inside a single request handler. Expensive work is asynchronous, extractor implementations are replaceable, and persistence/storage concerns are separated from the HTTP layer.

Author

Dovlet Aydogdyyev — Senior Software Engineer
Python · Django · FastAPI · Flutter · React