Robot Manipulation & Control Intern

Realhand Inc.; Palo Alto (CA); Hybride; USD 44 000 - 103 000; Temps partiel

Résumé du poste

RealHand Inc. in Palo Alto, CA is seeking a Robot Manipulation & Control Intern to work on controlling dexterous hands and arms, building controllers, perception stacks, and learning-based policies for real hardware.

You will collaborate with ML teams, simulate with physics engines, and write production-quality Python/C++ code for robotic platforms. The internship focuses on translating policy into action with emphasis on real-world validation in simulation and on hardware, with a direct path to

Qualifications

• Experience deploying software on real robot hardware.
• Strong robotics fundamentals: kinematics, dynamics, control theory, motion planning, and contact mechanics.
• Experience with ROS/ROS2 and full robotics software stacks.
• Experience with physics simulators for contact-rich manipulation (Isaac Sim, MuJoCo, Genesis, SAPIEN, etc.).
• Familiarity with egocentric and wrist-mounted cameras; calibration challenges.
• Proficiency in Python and C++; experience with a DL framework.

Responsabilités

• Develop and tune controllers for dexterous hands and arms: impedance and admittance control, force and tactile feedback, compliant contact behavior, and in-hand manipulation primitives
• Implement manipulation behaviors including grasping, pick-and-place, object reorientation, tool use, and bimanual coordination — as classical pipelines, learned policies, or hybrids
• Build perception for manipulation: egocentric and eye-in-hand pose estimation, grasp affordance prediction, 6-DoF object pose, and visual servoing under occlusion by the hand itself
• Model the hand and its contact behavior in physical simulation, validate that simulated controller performance predicts real performance, and close the gap where it does not
• Write and maintain production-quality Python and C++ for robot platforms: real-time loops, hardware interfaces, safety limits, and tests
• Collaborate with the ML team to deploy learned policies on hardware — latency budgets, observation and action space design, safety envelopes, and failure fallback
• Build calibration, system identification, and diagnostic tooling so the platform stays trustworthy over months of use
• Evaluate tradeoffs between classical control and learning-based approaches and make principled design decisions
• Continuously improve robotic software for performance, reliability, and scalability

Connaissances

Robotics fundamentals

Python

C++

ROS/ROS2

Control theory

Experiment design

Simulation experience

Formation

MS/PhD in Robotics or CS or related

Outils

Isaac Sim

MuJoCo

Genesis

SAPIEN

Robot hardware interfaces

Description du poste

# Robot Manipulation & Control InternPalo Alto, CAInternshipOn-site$20-$75/hour based on experiencePosted July 30, 2026Apply for this roleAbout the role## OverviewRealHand is an AI robotics company building dexterous robotic hands and the intelligence that drives them. The goal of the company is to give machines human-level dexterity — hardware that survives real-world contact, and policies learned from how humans actually use their hands. Our hands are deployed on humanoid and mobile platforms across research, automation, and human-robot collaboration. RealHand is headquartered in Palo Alto, CA.We are looking for a Robot Manipulation & Control Intern to work on the layer where a policy's intent becomes force on an object. You will write controllers, planners, and perception stacks for our dexterous hands and arm platforms, deploy state-of-the-art learning-based methods on real setups, and be responsible for whether they actually work day after day. Contact-rich manipulation punishes sloppy physics, so you will spend real time in simulation as well as on hardware — and care about the gap between them.This is an on-site internship with a direct path to a full-time role.Salary range: $20-$75/hour based on experience.What you will do## Responsibilities* Develop and tune controllers for dexterous hands and arms: impedance and admittance control, force and tactile feedback, compliant contact behavior, and in-hand manipulation primitives* Implement manipulation behaviors including grasping, pick-and-place, object reorientation, tool use, and bimanual coordination — as classical pipelines, learned policies, or hybrids* Build perception for manipulation: egocentric and eye-in-hand pose estimation, grasp affordance prediction, 6-DoF object pose, and visual servoing under occlusion by the hand itself* Model the hand and its contact behavior in physical simulation, validate that simulated controller performance predicts real performance, and close the gap where it does not* Write and maintain production-quality Python and C++ for robot platforms: real-time loops, hardware interfaces, safety limits, and tests* Collaborate with the ML team to deploy learned policies on hardware — latency budgets, observation and action space design, safety envelopes, and failure fallback* Build calibration, system identification, and diagnostic tooling so the platform stays trustworthy over months of use* Evaluate tradeoffs between classical control and learning-based approaches and make principled design decisions* Continuously improve robotic software for performance, reliability, and scalabilityWhat you bring## Requirements* Demonstrated experience deploying software on real robot hardware* Strong robotics fundamentals: kinematics, dynamics, control theory, motion planning, and contact mechanics* Experience writing controllers and integrating them with a full robotics software stack (ROS/ROS2 or equivalent middleware)* Practical experience with a physics simulator for contact-rich manipulation (Isaac Sim, MuJoCo, Genesis, SAPIEN, or equivalent), and a clear-eyed view of where its contact model is and is not trustworthy* Familiarity with egocentric and wrist-mounted camera setups for manipulation, and the occlusion and calibration problems they create* Proficiency in Python and C++; experience with at least one deep learning framework* Ability to design experiments, analyze failures, and iterate quickly in real-world robotic systems* Thrive in fast-paced, ambiguous environments where solutions require exploration and ownershipNice to have## Bonus qualifications* Currently enrolled in an MS or PhD program in Robotics, Computer Science, EE, ME, or a related field* Experience with dexterous or multi-fingered hands, tactile sensing, or in-hand manipulation* Experience with imitation learning or RL for manipulation (diffusion policy, ACT, VLA models, or similar)* Teleoperation system design — gloves, exoskeletons, VR, or hand-tracking based* 3D geometry and multi-sensor fusion: RGB-D, stereo, LiDAR, point clouds, meshes* Publication record at ICRA, IROS, RSS, CoRL, or top-tier ML/CV venues* Passion for building robots that do real physical work outside the lab