Contributing

SIE is open-source under the Apache 2.0 license. Contributions are welcome.

Development Setup

Prerequisites

• Python 3.12
• uv
• Git

Clone and Install

The repository root is a virtual Python workspace. From the repository root, install and verify every workspace member with the committed lock:

git clone https://github.com/superlinked/sie.git
cd sie

uv python install 3.12
uv lock --check
uv sync --frozen --all-packages

Project Structure

sie/
├── packages/
│   ├── sie_sdk/            # Python SDK
│   ├── sie_server/         # GPU inference server
│   ├── sie_gateway/        # Rust inference gateway
│   ├── sie_config/         # Config control plane
│   ├── sie_server_sidecar/ # Rust worker sidecar
│   ├── sie_telemetry/      # Telemetry facade
│   ├── sie_audio_prep/     # Audio preprocessing
│   ├── sie_mcp/            # MCP server
│   └── sie_ts_sdk/         # TypeScript SDK
├── integrations/           # Framework and vector-store adapters
│   ├── sie_langchain/      # LangChain adapter
│   ├── sie_llamaindex/     # LlamaIndex adapter
│   ├── sie_chroma/         # Chroma adapter
│   └── ...                 # See integrations/ for the full set
├── deploy/
│   ├── helm/               # Kubernetes Helm charts
│   └── k8s/                # Raw Kubernetes manifests
└── examples/               # Example apps and notebooks

---

Running Tests

Tests run via pytest against the committed lock:

uv run --frozen --project . --no-sync pytest -c pyproject.toml

Test Types

Type	Requires Server	When to Run
Unit tests	No	Every PR
Integration tests	Yes	When changing SDK/server interface
Docker tests	Docker	When changing Dockerfiles
GPU tests	GPU worker	When changing inference pipeline

---

Code Style

Formatting and linting use ruff, and type checking uses ty; both are configured in the root pyproject.toml. SIE uses standard Python formatting and type annotations throughout.

---

Adding a New Model

See Adding Models for the full guide. The short version:

1. Create a YAML config in packages/sie_server/models/
2. Map to an existing adapter (or write a new one in packages/sie_server/src/sie_server/adapters/)
3. Add the model’s task declarations to its config YAML
4. Submit a PR

---

Adding a New Integration

1. Create the package directory under integrations/:

   integrations/sie_myframework/
   ├── src/sie_myframework/
   │   ├── __init__.py
   │   └── embeddings.py
   ├── tests/
   │   ├── test_embeddings.py
   │   └── test_integration.py
   └── pyproject.toml

2. Implement the framework primitives:

   • encode() → framework’s embedding interface
   • score() → framework’s reranker/compressor interface
   • extract() → framework’s tool/extractor interface

3. Use shared test fixtures from integrations/conftest.py:

   • mock_sie_client - mocked SIEClient for unit tests
   • sie_server_url - real server URL for integration tests

4. Add tests and submit a PR

---

PR Workflow

1. Fork the repository
2. Create a feature branch: git checkout -b feature/my-change
3. Make your changes with tests
4. Run uv run --frozen --project . --no-sync pytest -c pyproject.toml
5. Submit a pull request

---

License

SIE is licensed under Apache 2.0. By contributing, you agree that your contributions will be licensed under the same terms.