Review the evidence signals before interviewing. Then use the anchored descriptions—not instinct alone—to choose the score that best matches each answer.
01
Evaluation factor
Method and rigour
35% weight
Check command of hazard analysis methods: HARA, STPA, FMEA, fault tree analysis, and how they applied ISO 26262 ASIL decomposition or ISO 21448 SOTIF triggering condition analysis.
Evidence to listen for
Follows and can justify an established methodology
Understands contamination, bias, and chain of custody as they apply to the field
Knows the limits of their techniques and says so
Documents procedure so results are reproducible and defensible
Five-point scoring guide
1
Poor
Careless method; unaware of contamination, bias, or procedural integrity.
2
Needs Improvement
Knows procedures but applies them inconsistently; gaps in documentation.
3
Satisfactory
Sound standard practice; less certain outside familiar techniques.
4
Very Good
Rigorous and well documented; understands the limits of each method.
5
Excellent
Names the specific method used per hazard, explains ASIL or SOTIF rationale, and cites acceptance criteria they helped define.
02
Evaluation factor
Real casework
25% weight
Probe actual work on a driving automation programme: disengagement and collision triage, ODD definition, scenario libraries in CARLA or resimulation, safety case evidence they authored.
Evidence to listen for
Brings specific cases, sites, or projects rather than general description
States their own role and what they personally handled
Can describe an ambiguous or degraded case and how they proceeded
Knows what happened to the work afterwards
Five-point scoring guide
1
Poor
No hands-on casework; experience is entirely academic.
2
Needs Improvement
Limited exposure; cannot describe their own contribution clearly.
3
Satisfactory
Real casework with adequate detail; ownership sometimes vague.
4
Very Good
Specific cases with clear personal scope and outcomes.
5
Excellent
Describes real incidents triaged, fleet miles or scenario counts reviewed, and named artefacts such as UL 4600 safety case chapters.
03
Evaluation factor
Interpretation and judgement
25% weight
Test how they reason from sparse event data: distinguishing perception misdetection from planner error, judging residual risk, deciding whether a release gate should hold.
Evidence to listen for
Separates what the evidence shows from what they infer
States confidence levels and what would change their conclusion
Comfortable saying the result is inconclusive
Handles pressure to reach a preferred conclusion without bending
Five-point scoring guide
1
Poor
Overstates findings; no separation of evidence from inference.
2
Needs Improvement
Reaches conclusions the evidence does not support; uneasy with uncertainty.
3
Satisfactory
Reasonable judgement; qualifies findings when prompted.
4
Very Good
Clearly separates evidence from inference and states confidence unprompted.
5
Excellent
Walks through a specific ambiguous event, states the competing hypotheses, and explains the evidence that settled the release decision.
04
Evaluation factor
Reporting and testimony
15% weight
Assess written and verbal reporting to engineering leads, regulators, and NHTSA or state DMV submissions; ask for examples of safety review board presentations or corrective action records.
Evidence to listen for
Writes findings that a non-specialist can act on
Has presented or defended work to an external audience: court, client, review board, publication
Withstands challenge without overclaiming or retreating
Keeps records that hold up to scrutiny
Five-point scoring guide
1
Poor
Cannot communicate findings; records would not withstand review.
2
Needs Improvement
Reporting is unclear or incomplete; avoids external scrutiny.
3
Satisfactory
Adequate reports; limited experience defending work externally.
4
Very Good
Clear reporting and real experience presenting to an external audience.
5
Excellent
Produces clear, traceable findings; has defended conclusions to a safety board or regulator without softening the risk.
Put this rubric to work
Score every candidate against the same standard
Add these weighted factors to Hirevire and let AI evaluate recorded answers against your rubric.
