Senior Machine Learning Engineer, End‑to‑end Autonomous Driving

NVIDIA NVIDIA · Semiconductors · Santa Clara, CA

Senior Machine Learning Engineer at NVIDIA focused on building, training, and deploying large-scale end-to-end autonomous driving models using VLM/VLA architectures and a data flywheel for continuous improvement. The role involves designing models, driving data collection and iteration, curating multimodal datasets, developing data-centric algorithms, exploring new data sources, and creating agentic data workflows.

What you'd actually do

  1. Designing, implementing, and training large‑scale end‑to‑end driving models.
  2. Driving the data flywheel: identifying failure cases, specifying data collection and labeling needs, and iterating models to close real‑world performance gaps.
  3. Building, curating, and maintaining high‑quality multimodal datasets (e.g., video, sensor, language/action traces) tailored for end‑to‑end autonomous driving.
  4. Developing and applying data‑centric learning algorithms such as active learning, curriculum learning, automated hard‑example mining, outlier and novelty detection, and semi/self‑supervised methods.
  5. Exploring and productizing new data sources including simulation, synthetic data, and world‑model‑based generation/augmentation to improve coverage and robustness.

Skills

Required

  • Python
  • PyTorch, TensorFlow, or JAX
  • Deep learning
  • Transformer-based architectures
  • Video modeling
  • Multimodal VLM/VLA or foundation models
  • Data preprocessing
  • Distributed training
  • Evaluation
  • Debugging
  • Iterative improvement
  • Active learning
  • Curriculum learning
  • Outlier/novelty detection
  • Large-scale sample mining
  • Software engineering practices
  • Testing
  • Code review
  • CI/CD
  • Collaboration
  • Communication

Nice to have

  • Experience building and operating data flywheels or large‑scale data pipelines for ML
  • Data quality monitoring
  • Continuous retraining loops
  • End-to-end driving models
  • Large-scale behavior cloning
  • Reinforcement/imitation learning for driving or robotics
  • Simulation
  • Synthetic data
  • World models
  • Impactful publications
  • Open-source projects
  • Safety, reliability, and validation requirements for autonomous driving or other safety-critical applications

What the JD emphasized

  • PhD with 4+ years, MS with 6+ years, or BS (or equivalent experience) with 8+ years of relevant experience
  • Hands-on experience training and deploying deep learning models on real‑world datasets
  • Track record of leading complex cross‑team projects, setting technical direction, and making critical technical decisions that impact multiple teams or products.

Other signals

  • end-to-end autonomous driving models
  • VLM/VLA architectures
  • data flywheel
  • large-scale training and deployment
  • data-centric learning algorithms