V1/development ai ml

[Aytsuu]

This pull request introduces a new "LoopHarvest AI Engine" stack, providing a local AI API gateway using FastAPI and Ollama, containerized with Docker Compose. The implementation includes a secure gateway service, configuration and environment management, model and error schemas, utilities, and a comprehensive test suite.

Core service and infrastructure:

• Added a docker-compose.yml file to orchestrate the Ollama model server and a FastAPI-based gateway, exposing ports and configuring volumes and dependencies.
• Introduced a README.md with setup instructions, project structure, and usage guidance for the new AI engine stack.
• Created a Dockerfile for the gateway service, installing dependencies and launching the FastAPI app with Uvicorn.
• Added pyproject.toml and requirements.txt for dependency and build management, including FastAPI, HTTPX, and testing tools. [1] [2]
• Provided an .env.example file to document required environment variables for configuration.

Gateway application logic:

• Implemented the FastAPI gateway in src/main.py, including endpoints for health checks, model listing, and chat requests, with dependency injection, error handling, and upstream communication to Ollama.
• Added configuration management via src/config.py using Pydantic settings, supporting environment-based overrides.
• Enforced API key authentication with dependency logic in src/dependencies.py.
• Defined Pydantic models for chat requests, responses, and error envelopes in src/models.py.
• Added a utility in src/utils/image.py for handling image sources and media types (not yet integrated into endpoints).

Testing and quality:

• Added fixtures and configuration for pytest-based testing in tests/conftest.py.
• Implemented a test suite in tests/test_gateway.py with mock upstreams to verify endpoints, authentication, error handling, and chat functionality.

These changes together establish a robust, testable, and extensible local AI API stack for LoopHarvest.