Dotscience docs

Welcome to the Dotscience documentation

Dotscience is an enterprise-grade deployment and collaboration platform for machine learning, that makes it easy to take your data science process into production via MLOps (DevOps for Machine Learning).

Dotscience works by arranging your team's data science work into projects, storing project metadata in the Dotscience hub. The hub is either ours on the cloud, or yours on-premise, meaning we can be run on-cloud, on-premise, or hybrid. The user supplies their own compute via either cloud instances or their own machines, known in our system as runners. To interact with our system, either sign in / sign up and use our GUI + SaaS (e.g., the built-in Python JupyterLab), or install to your own system and use scripts or command line. All work within the system is automatically versioned and recorded, including code, datasets, data provenance, models, model parameters, notebooks, and runs. This ensures that everything is fully accountable and reproducible. Deployment to production and statistical monitoring are simplified via our built-in use of Docker, Kubernetes, Prometheus, and Grafana, meaning full production can be achieved without requiring infrastructure setup. Or you can configure your own using our API, endpoints, etc.

For more information, navigate to the subsections below...