AI Project Template

A lightweight, modern Python project template used by the AI Service Center Berlin-Brandenburg (HPI) for AI project repositories.

and provides a clean, standardized starting point for AI projects by the AI Service Center Berlin-Brandenburg.

Creating a New Project

Use uvinit with the AI HPI template:

uvx uvinit --template gh:aihpi/template-ai-project

Then, just follow the instructions.

What is this?

This repository provides a standardized AI project scaffold designed specifically for AI and research-oriented development at AI Service Center Berlin-Brandenburg (HPI).

The goal is to give AI@HPI teams a consistent, maintainable, and modern baseline for building AI Python applications, prototypes, research code, and AI-related utilities, without unnecessary complexity.

The template is based on the excellent simple-modern-uv template by jlevy.

What this template includes

The template offers:

• uv for fast dependency management, virtual environments, packaging, and builds
• Automatic versioning derived from Git tags, removing the need for manual version bumps.
• ruff for linting and code formatting
• BasedPyright for static type checking
• pytest with pytest-sugar for testing
• codespell for spell checking
• Preconfigured GitHub Actions workflows:
  • Continuous Integration (lint + type check + tests)
  • Package building and optional publish-to-PyPI workflow
• Generated projects include modular, editable docs:
  • installation.md – how to set up and install dependencies
  • development.md – development workflows (tests, linting, formatting, etc.)
  • publishing.md – how to publish releases (if relevant)

What this template doesn't include

This template doesn't include:

• Domain-specific AI frameworks or dependencies (e.g., PyTorch, TensorFlow, Hugging Face)
• Integration with AI-assistant IDE rules.
• Data pipelines or preprocessing scripts
• Complex deployment tooling (Docker, Kubernetes, cloud CI/CD)
• Detailed documentation templates beyond basic installation.md, development.md, and publishing.md
• Advanced project organization for very large repositories (e.g., multi-package monorepos)

What do we want to add to this template

For improving the template and further streamline AI HPI projects, we are considering possible enhancements including:

• Optional integration with ML/DL frameworks (PyTorch, TensorFlow) and environment configurations for GPU support
• Prebuilt data handling and preprocessing utilities for common AI workflows
• Template notebooks or example scripts for standard tasks such as model training, evaluation, and visualization
• Optional Dockerfile and cloud deployment templates for easier productionization
• A standardized logging and experiment tracking setup (e.g., with MLflow or Weights & Biases)
• Extended documentation templates for research papers, reproducibility, and project reporting
• Template type stubs and mypy integration for stricter type checking in AI-heavy projects

About AI HPI / KI-Servicezentrum

The AI Service Center Berlin-Brandenburg (HPI) supports organizations in adopting state-of-the-art AI through research, compute resources, consulting, and educational offerings. More information: https://hpi.de/ki-servicezentrum/