Senior AI/ML Engineer

General Motors

. Validate optimized implementations against reference implementations by defining equivalence and designing inputs that expose violations .

Posted 10/21/2026full-timeSunnyvale • California • United StatesSenior💰 $170,600 - $261,300 per yearWebsite

Core Competencies

Role fit

Core Competencies

Use this summary to align your resume positioning with the role.

Demonstrates expertise in numerical analysis, matrix theory, and neural network training dynamics, with a strong focus on building analytical tools for production training and evaluation. Proficient in PyTorch and Python, with hands-on experience in debugging large-scale training runs and optimizing distributed training processes.

Highest-signal resume keywords

Numerical AnalysisJacobian/Hessian EstimationNeural Network Training DynamicsPyTorchDistributed Training

ATS Keywords

Tailor your resume

Applicant Tracking System Keywords

Tip: use these terms in your resume and cover letter to boost ATS matches.

Hard Skills

Numerical AnalysisMatrix TheoryJacobian EstimationHessian EstimationGradient PropagationLoss Landscape AnalysisQuantizationCompiler ToolchainsDebugging Large-Scale Training RunsBuilding Analytical Tools

Tools & Technologies

PyTorchPythonFSDPMegatron-LMDeepSpeed3D Parallelism

Certifications & Qualifications

Bachelor's DegreeMaster's DegreePhD

Industry Keywords

Model RobustnessOOD GeneralizationAdversarial AnalysisADAS SystemsEvaluation Pipelines

Tech Stack

Tools & technologies

PythonPyTorch

About the role

Key responsibilities & impact

Validate optimized implementations against reference implementations by defining equivalence and designing inputs that expose violations
Connect numerical differences from quantization, compilation, and precision reduction to downstream driving behavior
Analyze open-loop metrics such as trajectory displacement error and perception IoU
Evaluate closed-loop outcomes and identify mechanisms behind behavioral disparities
Build Jacobian/Hessian-based sensitivity and robustness analysis tooling
Characterize model responses to weight and input perturbations, including out-of-distribution inputs
Create tooling that runs repeatedly across checkpoints for use by other engineers
Design diagnostics to detect and root-cause training instabilities, including gradient vanishing and explosion, loss spikes, and silent divergence
Decompose gradient and update trajectories into loss-descent and oscillatory components under schedules such as WSD
Optimize metric computation, gradient decomposition, and diagnostic logging for distributed training with negligible throughput cost and no OOM risk
Define numerical thresholds that support ship/no-ship deployment decisions for compressed models

Requirements

What you’ll need

Working command of numerical analysis and matrix theory
Experience with Jacobian/Hessian estimation, spectral properties, conditioning, and floating-point error analysis
Strong mental model of neural network training, including loss landscapes, gradient and error propagation, optimizer dynamics, and training failure modes
Adversarial approach to numerical edge cases, including denormals, extreme dynamic range, catastrophic cancellation, degenerate shapes, and accumulation-order effects
High proficiency in PyTorch and Python
Track record of building mathematically defensible analytical tools fast enough for production training and evaluation loops
Ability to make numerical analysis actionable and trace anomalous results to their source operations
Bachelor's, Master's, or PhD in Applied Mathematics, Control, Physics, Computer Science, Data Science, or a closely related quantitative field
Hands-on experience debugging large-scale training runs, diagnosing loss spikes, resolving numerical divergence, and investigating training dynamics is a competitive edge
Experience with distributed training beyond DDP, including FSDP, Megatron-LM, DeepSpeed, or 3D parallelism, is a competitive edge
Experience in quantization, compiler toolchains, or inference-time numerical parity is a competitive edge
Experience building or scaling evaluation pipelines and metrics for AV or ADAS systems is a competitive edge
Published or applied work in model robustness, OOD generalization, or adversarial/perturbation analysis is a competitive edge
Selected candidate must report to the specified location at least 3 times per week
Willingness to travel less than 25%

Benefits

Comp & perks

Bonus potential through an incentive pay program based on company performance, job level, and individual performance
Medical, dental, and vision benefits
Health Savings Account
Flexible Spending Accounts
Retirement savings plan
Sickness and accident benefits
Life insurance
Paid vacation and holidays
Relocation benefits for candidates who qualify under company policy