2026 Summer Intern, Machine Learning Engineer, Trajectory Generation (PhD)

General Motors•Sunnyvale, CA

15d•Hybrid

About The Position

About the Team: The trajectory generation team (TGX) is responsible for turning scene understanding into wholistic AV behavior. The team must balance multiple competing objectives. Trajectories need to be comfortable for passengers (minimizing jerk and acceleration), obey traffic laws, respect vehicle dynamics constraints (steering limits, acceleration capabilities), avoid obstacles, and respond in real-time to changing road conditions. The TGX team uses state-of-the-art ML, RL, and optimization techniques to improve the GM driving product and support the objectives above. About The Role: As an intern on the trajectory generation team, you would work on well-scoped research or engineering projects that contribute to the team's ML-driven planning stack. This might involve implementing and benchmarking new neural network architectures for behavior prediction, improving data pipelines for training trajectory models, or conducting ablation studies on different learning approaches. The intern would collaborate closely with senior engineers and researchers, participating in code reviews, team meetings, and design discussions while gaining hands-on experience with production autonomous driving systems. They'd work with real-world driving data, simulation environments, and potentially see their contributions deployed to test vehicles. The role offers a unique opportunity to bridge cutting-edge ML research with safety-critical robotics applications. An intern would develop skills in deep learning frameworks (PyTorch), work with large-scale distributed training systems, and learn how to validate and verify learned models for autonomous driving. Beyond technical skills, they'd gain insight into the challenges of building reliable AI systems for the real world—handling edge cases, ensuring safety guarantees, and balancing model performance with computational constraints. The experience provides exposure to how ML teams operate within a complex, multi-disciplinary engineering organization building toward full autonomy.

Requirements

Currently enrolled in a PhD program in Computer Science, Machine Learning, Robotics, or a related STEM field.
Hands-on experience with one or more machine learning frameworks (e.g., PyTorch, TensorFlow, JAX, or Keras).
Availability to work full-time (40 hours per week) during the internship period.
Demonstrated coursework, research., or projects in AI/ML.
Strong programming skills in Python.

Nice To Haves

Exposure to deep learning architectures such as Transformers, CNNs, or Diffusion Models.
Exposure to deep reinforcement learning algorithms such as PPO, DQL, and distributional adaptations.
Experience with robotics through projects or research.
Familiarity with multimodal learning or working with complex, temporal data.
Interest in contributing to publications, open-source projects, or patents.
Familiarity with systems programming languages (e.g., C++ or Java) is a plus.
Intent to return to degree-program after the completion of the internship.
Graduating between December 2026 and June 2027.

Responsibilities

Implement and experiment with neural network architectures for trajectory prediction, behavior planning, or mission planning tasks
Build data pipelines and visualization tools to process, analyze, and evaluate large-scale driving datasets
Train and benchmark ML models using distributed compute infrastructure, running ablation studies to optimize performance
Validate models in simulation environments and analyze failure cases across diverse driving scenarios
Participate in code reviews, team meetings, and technical discussions while documenting experiments and results
Contribute to production codebases by prototyping new approaches for handling challenging driving situations like merges or complex intersections