Drimva

Atoms

Safety Validation Lead, Scenarios

$182K–$231K

San Francisco, CA +1 · Full-time · Posted Sep 1, 2026

Location
San Francisco, CA
San Francisco, California, United States
Experience
8+ years in safety validation, verification, or test engineering for automated vehicles, ADAS, or another safety-critical domain.

About the job

Owns the safety-relevant scenario catalog, tooling, and coverage argument used to validate automated-vehicle safety, drawing on fleet logs and published crash-avoidance protocols.

What you'll do

  • Define scenario relevance and coverage for operating domains; maintain a coherent catalog and defend its sufficiency.
  • Build and maintain tools for scenario extraction, coverage queries, catalog freshness, and regression testing.
  • Own log-based scenario extraction and criticality measures, adapt standardized crash-avoidance suites, and define what scenario results can support.

What they ask for

  • Hands-on experience with ASAM OpenSCENARIO and OpenDRIVE, ISO 21448 (SOTIF), and ISO 34502.
  • Strong Python skills and direct experience working with large volumes of vehicle log data; built tooling used by other engineers.
  • Hands-on experience with crash-avoidance portions of published consumer test protocols, such as Euro NCAP's Crash Avoidance protocols in the 2026 series, the ADAS component of US NCAP, or equivalent.
  • Ability to distinguish claims, arguments, and evidence, and explain what results support.

Nice to have

  • Proving-ground or closed-course test execution, including instrumented targets and test equipment.
  • Experience with falsification, adversarial scenario search, criticality-driven parameter sampling, or contributing scenario-based evidence to a safety case.

Pay

$182,000 - $231,000 per year (base salary)

Tools

Python, ASAM OpenSCENARIO, OpenDRIVE, ISO 21448 (SOTIF), ISO 34502

AI focus

Use and shape AI tooling for safety analysis, including log-event classification and clustering, scenario variant generation and stress-testing, catalog gap checks, and mapping published protocols to the company taxonomy. Generated scenarios are hypotheses to verify, not proof of coverage; validation claims require human judgment and traceable justification.

Model-written summary · the original posting is authoritative

First seen Oct 29, 2026 · Last verified Nov 6, 2026 (UTC)