Dotscience

Dotscience set out to bring DevOps practice to machine learning. The profile held for it describes a vision in which AI must be reproducible, accountable, collaborative and continuously delivered for it to be safe and productive. Its current web address no longer appears to show this business, so this summary relies on the earlier profile.

What Dotscience does

• Reproducibility: tracking the many variables in machine learning so results can be repeated.
• Accountability: recording how models were made, which matters for compliance when models make critical decisions.
• Collaboration: helping ML teams work together and avoid confusion, rework and errors.
• Continuous delivery: retraining models and monitoring them for drift after release.

Key facts

• Focus: DevOps for machine learning
• Principles: reproducible, accountable, collaborative, continuous