At Reality Labs Research, our goal is to explore, innovate and design novel interfaces and hardware for the next generation of virtual, augmented, and mixed reality experiences. We are driving research towards a vision of natural, seamless experiences in XR environments that are effective, enjoyable, and functionally indistinguishable from those in the real world.
As a Research Scientist Intern, you will work at the intersection of representation learning, physics-based estimation, and robotic manipulation. Your primary focus will be on developing latent space representations of object physical state with learned, high-dimensional embeddings that can capture time-varying object properties during dexterous manipulation. You will design experiments to validate these representations and demonstrate their value in practical multi-sensory setup and applications.
Our internships are twelve (12) to twenty-four (24) weeks long and we have various start dates throughout the year.
Responsibilities
- Design and implement latent space representations for object physical state during robotics manipulation tasks that go beyond fixed parameter sets.
- Design and execute controlled experiments to validate the representation: measuring adaptation speed, property decoding fidelity, and downstream control performance against baselines (no object state, explicit physical parameters, raw sensor history).
- Benchmark the latent state representation on practical dexterous manipulation tasks.
- Collaborate with researchers and cross-functional partners including communicating research plans, progress, and results.
- Showcase the value in simulated or physical demos.
Minimum Qualifications
- Currently has or is in the process of obtaining a Ph.D. degree in Robotics, Machine Learning, Computer Science, or relevant technical field
- Strong background in representation learning, generative models, or neural implicit representations (e.g., Gaussian splatting, NeRF, structured latent variable models)
- Experience with physics-based estimation, state estimation, or system identification in robotic or dynamical systems (e.g. Bayesian filtering, online adaptation, or meta-learning for system identification)
- Experience with Python and PyTorch
- Must obtain work authorization in the country of employment at the time of hire and maintain ongoing work authorization during employment Experience with tactile sensing, force/torque sensors, or robot hand manipulation
- Familiarity with model-based control (MPC), reinforcement learning, or imitation learning for manipulation
- Experience with Bayesian filtering, online adaptation, or meta-learning for system identification
- Experience with experimental design and statistical evaluation of robotic systems
- Experience working and communicating cross functionally in a team environment
- Proven track record of achieving significant results as demonstrated by grants, fellowships, patents, as well as first-authored publications at leading workshops or conferences such as Robotics (RSS, ICRA, IROS, CoRL, T-RO, IJRR), Machine Learning (NeurIPS, ICML, ICLR, AAAI, JMLR), and Computer Vision (CVPR, ICCV, ECCV, TPAMI), or similar
- Intent to return to the degree program after the completion of the internship/co-op