Member of Technical Staff, Research
Bagel Labs is a physical AI research lab developing compact world-action models and distributed training systems for autonomous robot control.
Funding history
About Bagel Labs
Bagel Labs currently describes itself as a physical AI research lab focused on a General World Action Model for autonomous robot control across tasks, environments, and robot types. Its current work centers on PARIS, a distributed training architecture, and WorldDiT, a compact diffusion-transformer architecture that unifies world-state prediction with robot-action generation.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will set and pursue research directions in diffusion and flow models, world models, action representations, latent dynamics, and embodied policy learning. You will advance Distributed Diffusion Models for physical AI, design rigorous benchmark experiments, analyze failures, communicate findings clearly, and collaborate with systems engineers to keep experiments reproducible and scalable across heterogeneous compute.
Requirements
- Strong research judgment in diffusion, flow, score models, world models, representation learning, robotics, simulation, imitation learning, reinforcement learning, or video and multimodal generation.
- Experiment design skills, including baselines, metrics, and failure modes.
- Fluency in reading and implementing recent machine learning papers.
- Clear technical writing.
- Strong Python skills and experience with a modern machine learning framework.
Responsibilities
- Set and pursue research directions across diffusion, world models, action representations, latent dynamics, and embodied policy learning.
- Advance Distributed Diffusion Models for physical AI through distributed world models, expert ensembles, routing, specialization, and compositional generalization.
- Explore latent action representations for embodied AI.
- Design rigorous experiments on physical AI benchmarks with baselines, metrics, and failure analysis.
- Communicate results clearly to researchers and infrastructure engineers.
- Collaborate with systems engineers to keep experiments reproducible and scalable across heterogeneous compute.
Benefits
- Meaningful equity.
- Paid travel to top machine learning conferences worldwide.