Autonomy Researcher

About Menlo

We are working on embodied intelligence. Our mission is to scale general-purpose autonomy for real world problems, through large-scale learning, multi-modal data, and robust control. We build Asimov, an open-source humanoid robot platform, and the full software stack that powers it. We are looking for passionate researchers and scientists who thrive at the intersection of machine learning, robotics, and systems engineering, and want to see their research come alive in real robots

The Role

You will lead development of the algorithms and architectures that enable our robots to interact with and reason about the physical world. This role bridges world model research, embodied AI, and real-time robotics. You will design and advance the models and learning systems that power robotic agents through real world jobs.

What You'll Do

• Train and adapt large-scale VLAs and VLMs that predict multi-modal futures spanning video, proprioception, audio, and actions
• Develop systems for cross-modal grounding, enabling robots to interpret sensor data in context and build coherent world models
• Enable temporal reasoning and goal-directed behavior through hierarchical task decomposition and meta-reasoning
• Support human-robot interaction by recognizing intentions, interpreting social cues, and enabling collaborative workflows
• Deploy models into real-time humanoid and mobile robots
• Design and run evaluation pipelines to measure generalization and safety
• Collaborate with locomotion, simulation, and hardware teams to bridge sim-to-real transfer
• Publish and open-source datasets, models, and papers in parallel

What We're Looking For

• MS/PhD in Robotics, AI/Computer Science, or related field
• Proficiency in Python and C++, and deep learning frameworks (PyTorch / JAX)
• Deep research experience in GenAI, RL/IL, control, or multimodal learning
• Understanding of scaling laws, evaluation metrics, and training large models at scale
• Familiarity with real-robot systems, sensing, and embedded control integration
• Strong publication record or equivalent research output in relevant areas
• Familiarity with industry SOTA and latest research, e.g. Gr00t, Pi0, and related work

Bonus points for:

• Experience with transformer-based control policies or diffusion policy learning
• Work on humanoid locomotion, manipulation, or whole-body coordination
• Prior open-source contributions or published datasets in robotics or deep learning

Why Join Menlo?

You will be part of a tight-knit research team working at the frontier of embodied intelligence, where your work does not stop at publication. At Menlo, your models run on real hardware, your ideas ship fast, and your research directly shapes how Asimov reasons about and acts in the physical world. If you want genuine ownership over hard problems and the freedom to pursue them from first principles, this is the place.