Software Engineering Intern - Machine Learning Research, Perception (Spring 2027 Co-Op)

Rivian•Palo Alto, CA

9d•Onsite

About The Position

As a Machine Learning Research Intern on Rivian's Autonomy team, you will advance neural scene reconstruction and world modeling that turn real-world driving data into high-fidelity, re-renderable 3D environments. This research supports the development, testing, and validation of the machine learning models behind Rivian's autonomous driving system, with the goal of improving the speed and safety confidence with which autonomy features are delivered to customers. The Spring 2027 Co-Op runs from January through August 2027. Candidates must be enrolled as undergraduate or graduate students in an accredited program during the internship term, with an expected graduation date between December 2027 and 2029. This role requires full-time availability (40 hours per week) and onsite work for the entire program duration. The role specifically requires candidates to be pursuing a PhD.

Requirements

Currently pursuing a PhD degree at an accredited university.
Pursuing a degree in computer science, electrical engineering, statistics, or a closely related quantitative field.
Research background in one or more of 3D scene reconstruction, neural rendering or novel view synthesis, depth or geometry estimation, 3D scene understanding, or generative modeling.
Track record of first-author publications at top-tier machine learning or computer vision conferences and journals, such as CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR, SIGGRAPH, CoRL, or RSS.
Strong proficiency in Python and a modern deep learning framework, preferably PyTorch, with demonstrated ability to independently implement, train, and debug models.
Solid foundation in 3D vision mathematics, including multi-view geometry, camera models, coordinate transforms, and 3D representations such as point clouds, meshes, voxels, and implicit fields.
Experience applying 3D reconstruction, neural rendering, or world modeling to autonomous driving, robotics, or other embodied domains.
Experience with large-scale distributed training and multi-GPU or multi-node workflows.
Must be available to work onsite, full time (40 hours per week), for the entire January–August 2027 co-op.
Must be an undergraduate or graduate student in an accredited program during the internship term, with an expected graduation date between December 2027 and 2029.

Nice To Haves

Familiarity with simulation, sensor simulation, or synthetic data generation for training and evaluating machine learning systems.
Experience working with large multi-sensor datasets and their calibration, synchronization, and data-quality challenges.
Open-source contributions, released research code, or a strong public portfolio of 3D machine learning work.

Responsibilities

Research, prototype, and train 3D machine learning models for scene reconstruction, novel view synthesis, and neural rendering using multi-sensor driving data, including camera, lidar, radar, and vehicle pose.
Investigate and extend approaches such as neural radiance fields, 3D Gaussian splatting, feed-forward reconstruction, depth and geometry estimation, and generative or diffusion-based scene synthesis.
Design quantitative and qualitative evaluation methods for reconstruction and rendering quality, including photometric, geometric, and downstream task-driven metrics.
Scale training and inference pipelines to large real-world datasets using distributed, multi-GPU infrastructure.
Collaborate with perception, simulation, and infrastructure teams to integrate research prototypes into Rivian's internal research platform.
Present results regularly, contribute to internal technical documentation, and deliver a final presentation summarizing the research contribution.