Automated Driving Advanced Development Intern, Machine Learning Research

Toyota Research Institute•Los Altos, CA

5d•$45 - $65•Hybrid

About The Position

Toyota Research Institute (TRI) is seeking a Machine Learning Research Intern to join its autonomy team and help bring end-to-end machine learning models—from pixels to trajectories—into robust, testable, and deployable systems. The internship focuses on machine learning, systems engineering, and real-world deployment, contributing to the implementation, evaluation, and integration of ML components for perception, planning, and control, including simulation-based testing. Interns will collaborate with researchers, data engineers, and autonomy engineers to help models progress from prototype to production. The work is part of the Automated Driving Advanced Development division, which connects TRI research with Toyota products, services, and needs through partnership and collaboration. The division is working with Woven by Toyota on research and development of a fully end-to-end learned driving stack, in alignment with TRI robotics efforts in Diffusion Policy and Large Behavior Models. The posting describes this as a paid 12-week hybrid, in-office internship; it references both Summer 2027 and Summer 2026.

Requirements

Currently pursuing a Ph.D. or have equivalent experience in Computer Science, Robotics, Engineering, or a related field.
Proficiency in Python for implementing and evaluating research ideas.
Experience with machine learning frameworks such as PyTorch.
Understanding of version control, testing, and software engineering fundamentals.
Interest in collaborative engineering and building reliable machine learning systems for real-world autonomy.

Nice To Haves

Experience with ML engineering workflows, including data sampling and curation, preprocessing, model training, ablation studies, evaluation, deployment, and inference optimization.
Understanding of debugging and profiling on the NVIDIA CUDA stack.
Hands-on experience with metrics dashboards, experiment tracking, and ML operations tools such as Weights & Biases, MLflow, or Metaflow.
Hands-on experience with robotics or real-world sensor data such as video, lidar, IMU, or radar.
Experience with state-of-the-art architectures for object detection and 3D perception.
Familiarity with foundation models, pre-training and efficient fine-tuning, multimodal Transformer architectures, or large-scale distributed training.
Experience with ROS, simulation frameworks such as CARLA or Nvidia DriveSim, or vehicle interfaces.
Experience with robot motion planning techniques such as trajectory optimization, sampling-based planning, or model predictive control, or with automated driving domains such as perception, prediction, mapping, localization, planning, or simulation.
Relevant open-source contributions or technical project write-ups.

Responsibilities

Conduct research on generative modeling capabilities for end-to-end planning from vision in automated driving.
Implement scalable end-to-end architectures that process raw sensor data and generate vehicle trajectories, addressing long-tail driving scenarios with limited data coverage.
Prototype, validate, and iterate on model architectures using imitation learning and large-scale data, with a focus on robust performance across diverse scenarios.
Perform closed-loop evaluations in sensor simulations and real-world testing environments.
Explore multimodal and language-conditioned models, external data sources, and transfer learning to improve the applicability and generalization of end-to-end policies.