// loading
The same rubric grades every candidate, every scenario, every track. Accuracy carries the most weight because the ticket actually closing is the thing. Methodology and tool fluency follow; communication and real-world fit hold the corners. Nothing hidden, nothing weighted by vibe.
same rubric since 2024-09, changes are public, changes are versioned
= (3*30 + 3*25 + 3*20 + 2*15 + 3*10) / 3 = 88 of 100, pass, cert minted
We started at seven and dropped two: speed because it punished careful work, and code quality because IT is not software engineering. The remaining five each survive the test of being independently observable in a transcript, which is the only thing a grader can defensibly read.
A rubric is also a statement about what is not measured. Each item below was proposed at least once, debated, and rejected; usually because grading it would punish the wrong people, or because no transcript could support the call.
The first time I read the matrix I knew I was looking at a rubric a senior engineer wrote, not a product manager.
staff SRE, midwest insurance, 2025NYC Local Law 144 (AEDT) compliance: ITC ran an internal bias-audit pilot at the 4/5 threshold; full independent third-party audit is on the roadmap. The rubric is published and versioned so candidates and recruiters can read it.
3-minute scenario, graded the same way as a recruiter-paid one