Cssmerge Field Test Engineer at Cssmerge responsible for planning and executing scenario tests on proving grounds and public roads. Owns risk assessment, instrumentation, and test reporting.
Responsibilities
Plan and run scenario cases at proving grounds: translate a case specification into a test procedure, set up instrumented targets and VUT’s, define approach conditions and tolerance bands, and execute enough valid runs to support a result.
Run scenario cases on public roads with safety drivers, route and condition selection to actually encounter the target scenario, in-vehicle protocol, driver briefing, abort criteria, and disciplined recording of what occurred. Keep staged and opportunistic encounters clearly distinct, because they are not the same evidence.
Own the risk assessment for the testing itself: site layout, personnel positions, exclusion zones, emergency stop provisions, weather and surface limits, and the conditions under which a test does not run. You hold the call to stop, and we expect you to use it.
Own instrumentation and ground truth - ego positioning, target position and velocity, timing synchronisation, video and data acquisition. A run that cannot be reconstructed from its data did not produce evidence.
Own tolerance bands, the number of valid runs a result requires, and the rejection of runs that fall outside them, with the reason recorded. Nobody downstream can audit this if you do not.
Write test reports a safety case engineer can cite directly: what was run, under what conditions, against what criteria, with what outcome, and every deviation from the procedure. Write for someone who was not there and will be asked to defend the conclusion.
Own proving ground scheduling, test equipment and its calibration, and third-party test service relationships, including the practical knowledge of what each venue can and cannot stage.
Report what is not testable as specified, what needed adaptation to run at all, and what no available venue can stage. A case definition nobody can execute is a defect in the definition, and you find it first.
Use AI tooling to remove the drudgery: first-draft procedures generated from case specifications, automated validity checking of runs against tolerance bands so invalid runs are caught at the track rather than in analysis a week later, and report drafting from acquisition data.
Qualification
Hands-on with the equipmentRigour about validitySafety-of-test judgmentData ability in Python or similarClear written communication
Required
5+ years hands-on vehicle test engineering, ADAS or AV testing, proving ground operations, or OEM and Tier 1 vehicle test. Time actually running tests, not planning them for other people.
Hands-on with the equipment: instrumented targets including soft cars and pedestrian and cyclist targets, driving robots, RTK or GNSS ground truth, timing synchronisation and data acquisition. Tell us what you have run, on what, and what went wrong with it.
Experience executing published test procedures, Euro NCAP or NHTSA crash-avoidance procedures or equivalent, and the judgment to see where a procedure written for a human-driven vehicle needs adaptation.
Rigour about validity. You reject runs that fall outside tolerance and record why, without being asked and without being talked out of it. This is the trait that most predicts success here.
Safety-of-test judgment. You have stopped a test, or refused to start one, and can explain the call. Staging hazards deliberately is the job; doing it without hurting anyone is the skill.
Data ability in Python or similar, comfortable reducing acquisition data and producing a defensible result rather than handing over raw logs.
Comfort using AI tooling to automate procedure drafting, run validity checks and report generation, with a clear-eyed view of the limit: a run either happened as specified or it did not, and no tool can decide a marginal run counts.
Clear written communication. Your reports become cited evidence in arguments other people sign, and a report that cannot be followed by someone who was not present is not usable.
Willingness to travel to proving grounds and test routes regularly, and to work the hours that test windows and weather actually allow.
