AI proposed a scoring rubric per job description. Nothing is scored until a human approves — this is the fairness gate. Each rubric defines which artifacts count, how they're weighted, and what good work looks like for that role.
2 rubrics pending review. Scoring will not begin for unapproved roles.
Generated from the AI Engineer JD in Odoo. Weights tuned toward reliability given platform-hardening quarter.
Does the change do what it claims, with evidence (tests, validation)?
Readable, well-scoped, maintainable; blast radius matches intent.
Moves the system toward fewer incidents and safer failure modes.
Dual-role rubric. Manager is scored on team leverage AND personal craft.
Removes blockers and shortens cycle time for reports.
Decisions align team capacity with weighted goals.
Personal contributions stay sharp where they matter.
Draft — awaiting director sign-off in the decision queue.
Capacity is invested in proportion to goal weight.
Cross-unit risks are surfaced and acted on early.
Draft — pending Design Lead approval.
Reduces friction in the user's real task.
Visual + interaction polish on its own terms.
Buildable within engineering constraints.