Watch the work
A React workspace brings model health, typed workflow editing, prediction records, and experiment comparison into view. Raw inputs and their results remain linked.
04 / ML operations
GigaMaid ML
Keep your production models in line.
The mascot is ridiculous. Quietly replacing a production model because a chart looks worrying would be worse. GigaMaid gives ML workflows a control plane with evidence, policy, and a human decision before release.
Predictions keep running while the system gathers feedback and evaluates candidates. MLflow keeps ownership of experiments and artifacts; GigaMaid tracks the workflow and release decisions around them.
ML operations / Pre-1.0 alpha
No surprise promotions
Degradation opens an approval request. Candidate training, shadow evaluation, canary rollout, promotion, and rollback follow server-enforced transitions.
Still an alpha
FastAPI, Ray, MLflow, and an evidence ledger support local evaluation. The pre-1.0 project is intended for local or trusted-network use.
Inside the project.
Straight from the project.
Adapted from the project README. Explore the repository for the full story.
Have something in mind?