Essential session storage is required to sign in. Optional analytics remain off unless you allow them.
Take a simple ML model out of a notebook and make it reproducible and tracked. Version the CODE in Git and the DATA with DVC (or equivalent), pin the environment (requirements + a container), and fix random seeds. Convert training into a runnable script/pipeline that logs every run to an experiment tracker (MLflow Tracking or Weights & Biases) — parameters, metrics, and the model artifact. Demonstrate that a teammate could reproduce a specific run from its logged code+data version, and register the best model to a model registry (real API, e.g. mlflow.register_model).
Code (Git) + data (DVC) + environment versioned; seeds fixed; a logged run can be reproduced from its recorded versions.
Every run logs params/metrics/artifacts to a tracker; runs are comparable; the best model is registered via the real registry API.
Training is a runnable script/pipeline, not a manual notebook; parameterized.
No data/secrets in Git; clear README of how to run and reproduce.
Use this README structure to make the work reviewable.
DevPath fetches a read-only, size-capped snapshot. Submitted code is never executed.
# A reproducible, tracked training pipeline ## Goal Take a simple ML model out of a notebook and make it reproducible and tracked. Version the CODE in Git and the DATA with DVC (or equivalent), pin the environment (requirements + a container), and fix random seeds. Convert training into a runnable script/pipeline that logs every run to an experiment tracker (MLflow Tracking or Weights & Biases) — parameters, metrics, and the model artifact. Demonstrate that a teammate could reproduce a specific run from its logged code+data version, and register the best model to a model registry (real API, e.g. mlflow.register_model). ## Acceptance criteria - [ ] Reproducibility: Code (Git) + data (DVC) + environment versioned; seeds fixed; a logged run can be reproduced from its recorded versions. - [ ] Experiment tracking: Every run logs params/metrics/artifacts to a tracker; runs are comparable; the best model is registered via the real registry API. - [ ] From notebook to pipeline: Training is a runnable script/pipeline, not a manual notebook; parameterized. - [ ] Hygiene: No data/secrets in Git; clear README of how to run and reproduce. ## Verification - [ ] Document installation and run commands. - [ ] Record automated checks and their exact commands. - [ ] Add representative output, screenshots, or a short demo where useful. - [ ] Confirm failure paths and known constraints. ## Decision log ### Decision title - Context: - Choice: - Alternatives considered: - Tradeoffs: ## Evidence - Link each rubric criterion to the file, test, or artifact that demonstrates it. ## Known limitations - List what is intentionally out of scope and what should be improved next.