ML/AI Research Internship - PhD

Atoms • San Francisco, CA

1d • Onsite

About The Position

Atoms is seeking a PhD-level AI/ML Research Intern to connect advanced AI research with real-world physical actuation for autonomous transport platforms. The intern will contribute to research on World Models and Vision-Language-Action (VLA) models, translating new theoretical developments into scalable deployment on physical autonomous systems. The work supports how the platforms perceive, simulate, and navigate complex environments.

The role is based in the Mountain View or San Francisco office and is onsite five days per week. The hourly pay range is $70 per hour. Applications are accepted on an ongoing basis.

Requirements

• Currently enrolled in a PhD program in Computer Science, EECS, Robotics, Machine Learning, or a closely related quantitative field.
• Publications or submissions in top-tier machine learning or robotics venues, such as NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, RSS, ICRA, or CoRL.
• Graduate-level research background in at least one of: generative world models, video generation, or diffusion models; VLA models or embodied AI; reinforcement learning, trajectory optimization, or decision-making systems; or large-scale multimodal Transformer architectures applied to robotics or spatio-temporal data.
• Advanced hands-on proficiency with PyTorch or JAX for training and fine-tuning large-scale neural networks.
• Proficiency in Python and working knowledge of C++.

Nice To Haves

• Hands-on sensor-fusion experience integrating inputs from multi-camera setups, LiDAR, and Radar.
• Familiarity with Bird’s-Eye-View (BEV) visual representations, 3D occupancy forecasting, multi-task learning, or specialized data tokenization for spatial data.
• Experience with large-scale distributed training on GPU/TPU clusters using tools such as DeepSpeed, Megatron-LM, Ray, or PyTorch FSDP.
• Prior research or internship experience in autonomous driving, mobile robotics, or spatial AI.

Responsibilities

• Research and develop reinforcement learning and distillation techniques for trajectory planning.
• Integrate emerging research from the AI community and identify and prototype promising solutions.
• Design and prototype end-to-end multimodal models that translate real-time visual perception and high-level behavioral goals into physical vehicle actuation.
• Develop interactive world models from raw multi-sensor logs so the team can re-simulate events and query what a vehicle would see after changing its trajectory.
• Ensure autonomous driving models can adapt to novel urban environments and edge cases.
• Work with validation and research teams to evaluate models in simulated scenarios, detect regressions, and identify systemic performance bottlenecks.

What This Job Offers

• Career Level: Intern
• Education Level: Ph.D. or professional degree