The team is building the next generation of autonomous driving at the company, starting with eyes-off highway driving and moving toward point-to-point autonomy, with the aim of capability that is safe, reliable, comfortable and scalable across the vehicle lineup.
Autonomy Evaluation finds where the autonomous driver still falls short. Data from every mile driven on the road or in simulation feeds the work, which identifies what limits the system, how large each problem is and what fixing it would be worth.
The results guide what the autonomy teams build next and feed into software release decisions. The group works closely with Autonomy, Simulation, Systems, Safety and Software Validation.
Responsibilities
Own the evaluation of the autonomous driver, from overall driving behavior down to each subsystem.
Answer hard questions such as how to measure good driving beyond safety, which subsystem caused a failure and how much that error matters on the road, and how to surface failures nobody has thought to look for.
Turn those answers into methods that other teams adopt, priorities that other organizations commit to, and evidence that release decisions depend on.
Set multi-year technical direction for evaluation, choosing which questions matter, which methods to fund and which to retire.
Develop new ways to judge overall driving, including learned or VLM-based judges of driving quality, comparison with expert human drivers, counterfactual and closed-loop analysis, and discovery of long-tail failure modes.
Define how perception, prediction, planning and localization, including learned components, are evaluated, using metrics that show their real effect on driving rather than accuracy alone.
Trace poor driving back to the responsible subsystem or interaction and measure how subsystem errors combine into driving outcomes.
Define how to determine whether a software or model change truly improves driving, covering regression detection and version-to-version comparison, so engineering teams and release reviews get reliable answers quickly.
Turn observed behavior into measured, ranked problems by sizing them with metrics and scene classifiers, building a Pareto of the problem space and driving fixes with the owning teams.
Take responsibility for the scientific rigor of evaluation, including sampling, statistical confidence, metric validity and agreement between simulation and road results. State the assumptions and limits of each result and resolve the hardest disagreements between signals.
Convert results into clear recommendations for engineering and leadership covering what changed, how confident the team is and what comes next.
Mentor Staff and Senior engineers through research reviews, design reviews and written technical direction.
Write production Python and SQL with tests and code review, and prototype and ship code yourself.
Requirements
Have 10+ years of experience in autonomous systems, robotics, ML evaluation or applied data science, with a record of turning analysis into decisions and shipped improvements.
Have created evaluation or measurement methods that were adopted beyond your own team.
Have deep experience designing statistically sound evaluations of dynamic systems, including sampling, confidence intervals, experimental design, hypothesis testing and precision and recall of classifiers on time-series data.
Have strong production Python and SQL skills, including testing and code review.
Have a record of starting from raw observations, defining and sizing a problem and driving it to resolution with data others trust.
Can influence without authority, aligning peers, partner teams and leadership and securing commitments from organizations you do not manage.
Build trusted relationships with partners such as Autonomy, Simulation, Safety and Software Validation, understand their evaluation needs and stay engaged through ambiguity and disagreement.
Have grown senior engineers through mentorship, design reviews and written technical direction.
Turn ambiguous questions and uncertain results into clear recommendations for executive, engineering and cross-functional audiences.
Hold a Bachelor's, Master's or PhD in Computer Science, Robotics, Mechanical or Aerospace Engineering, Machine Learning, Statistics, Data Science or a related field, or have equivalent practical experience.
Preferred: experience in autonomous driving or field robotics, including large-scale analysis of driving logs and camera, lidar and radar data from simulation and on-road testing.
Preferred: experience producing evaluation evidence for validation and release decisions, ideally for driver-assistance or automated-driving products.
Preferred: experience applying VLMs or LLMs to event classification, triage or scenario discovery.
Preferred: experience with ML interpretability such as attribution, ablations or embedding analysis.
Preferred: strong data visualization skills that turn high-dimensional metrics into clear, trustworthy views.
Preferred: familiarity with C++, especially for reading or contributing to autonomy stack code.
Conditions
The base salary range is $250,000 to $384,600. The final offer within that range depends on factors relevant to the position.
The role is a senior individual contributor position with no direct reports.
Benefits
Health and wellbeing coverage includes medical, dental, vision, a Health Savings Account, Flexible Spending Accounts, a retirement savings plan, sickness and accident benefits, life insurance, paid vacation and holidays, tuition assistance, an employee assistance program and vehicle discounts.
Selected candidates who pass a motor vehicle report review become eligible for a company vehicle evaluation program, in which they are assigned a company vehicle to drive and evaluate.
Where you’d work
Fully remote
You can work from
United States
About the company
G
General Motors
Industry: Mobility
Also hiring in United States, Austin, United States, Sunnyvale, United States and 9 more places