1334 questions
Could not save this preference. Your choice applies only to this page.
Stage profile · Individual contributor
Completes bounded work with clear goals and guidance, understands the user and intended outcome, verifies the output, and communicates promptly.
Read these as observable expectations across repeated work, not as a checklist for a single project. First distinguish whether the role expects the work, whether the person had a chance to demonstrate it, and whether the evidence supports the claim. Lack of opportunity or evidence is not itself a negative judgment. AI use is judged by verified outcomes, sound boundaries, and the quality of decisions; choosing not to use AI can be sound.
These lenses and the A–H capability map answer different questions. One experience may provide evidence across lenses, but do not count the same result multiple times.
Understand the people served and intended outcomes, make evidence-based trade-offs, and verify contribution. For platform, infrastructure, reliability, and safety work, cost, efficiency, risk reduction, and internal-user outcomes count; direct revenue ownership is not universal.
Apply relevant knowledge to diagnose, design, implement or review, validate quality, and make evolution decisions. The professional leaves required depend on the role.
Help others work effectively through communication, coordination, review, shared methods, coaching, or organizational mechanisms. IC influence is not measured by direct reports; managers also own formal people responsibilities.
Observe questioning, reasoning, learning transfer, judgment updates, relationship handling, and recovery in work. This is not a fixed personality or potential verdict; do not infer it from extroversion, fluent speech, or willingness to overwork.
Use this structure to understand a real example. Strong judgment can appear in a failed project; a successful result alone does not prove personal causation.
The labels beside capability references describe that capability’s role in this standard; they do not make the whole standard optional. Any role or scenario condition is stated separately.
Evidence to inspect: Can restate the user, intended change, and completion conditions.
Evidence to inspect: Deliverable meets the agreement; risks or delays are raised early.
Applies when: Assess only when policy permits and the task is suitable for AI assistance; choosing not to use AI is not by itself negative evidence.
Evidence to inspect: Can explain the change and show execution or manual review evidence.
Evidence to inspect: Acceptance records connect feedback to follow-up changes.
Evidence to inspect: Work passes basic checks and key concepts can be explained.
Evidence to inspect: Can identify obvious failure paths or sensitive-data risks.
Applies when: Depends on role policy, task risk, and available tools; sound decisions not to use AI count as valid evidence.
Evidence to inspect: Changes include verification; generated content is not assumed correct.
Evidence to inspect: Examples explain state transitions and data constraints.
Evidence to inspect: Provides reproduction steps, eliminated hypotheses, and verification records.
Evidence to inspect: Collaborators know the status in time to act.
Evidence to inspect: Can explain why feedback was adopted or discussed.
Evidence to inspect: Does not make unauthorized commitments affecting others or users.
Evidence to inspect: Fewer repeat issues or reuse of a new method on a related task.
Evidence to inspect: States what observations support a view and what remains unverified.
Evidence to inspect: Corrects the record promptly and does not conceal issues or overstate results.
Evidence to inspect: Learning goals connect to visible improvement in later work.
Evidence to inspect: Explains support sought, recovery actions, and the outcome of another attempt.
