GRAM is a self replication company creating populations of insectoids for the physical economy.
Our first research frontier is self-preservation: the base case of physical self-replication. Our machines will survive, coordinate, and recover without humans. We believe scalable machine labor requires more than single-agent task generality or machines shaped in our image.
About the roleYou will build manipulation capabilities for insectoids operating across changing geometry, orientation, surface condition, and payload. The work spans task and motion planning, learned policies, contact reasoning, trajectory generation, force control, and deployment on physical hardware. You will determine where model-based, learned, foundation-model, and hybrid approaches perform best against measured results.
This is a staff-level individual-contributor role with authority over the manipulation architecture, cross-stack technical decisions, and measured manipulation-capability readiness. You will establish system boundaries and evaluation standards, lead design and failure reviews, mentor engineers, and remain hands-on in C++ and Python. Success is a manipulation system that completes defined physical tasks repeatedly, exposes why it failed, and improves against physical evidence rather than a scripted laboratory demonstration.
What you will do- Set the manipulation architecture and technical direction across task planning, sensing, learned-policy integration, contact reasoning, monitoring, recovery, and execution.
- Design manipulation behaviors for contact-rich tasks, including contact selection, constrained motion, force regulation, and recovery.
- Integrate and evaluate learned manipulation policies, including vision-language-action models and other embodied foundation models, through the full sensing, control, monitoring, and recovery stack.
- Implement production-quality planning and execution software in C++ and Python for real-time operation on insectoids.
- Lead cross-stack design reviews and resolve tradeoffs spanning sensing, control, actuation, operator interfaces, and mechanical constraints.
- Build task-level monitors that detect loss of contact, geometric mismatch, saturation, obstruction, and unsuccessful execution.
- Define hardware experiments, metrics, and regression tests for completion rate, cycle time, force accuracy, recovery rate, and failure mode.
- Analyze unsuccessful trials and convert the dominant causes into algorithm, sensing, controller, or mechanical changes that survive repeated physical test.
- Bachelor's degree in robotics, mechanical engineering, electrical engineering, computer science, or a related field, or equivalent practical experience.
- Demonstrated ownership of architecture and technical direction for a deployed manipulation system spanning planning, sensing, control, monitoring, and recovery.
- Strong C++ and Python skills, including experience with a robotics framework such as ROS 2, Drake, MoveIt, Pinocchio, or an equivalent internal stack.
- Built and deployed a closed-loop manipulation capability on a physical robot; you can explain its architecture and provide measured results from repeated trials.
- Led a failure-driven redesign that separated planning, perception, control, and hardware causes and produced a correction that survived repeated physical test.
- Whole-body manipulation, legged manipulation, mobile manipulation, or operation under changing orientation.
- Optimization-based planning, model predictive control, tactile sensing, compliance, or deployment of vision-language-action, diffusion, or other learned policies on physical robots.
- Setting technical standards, mentoring engineers, or shipping robotics software into field, manufacturing, aerospace, or other reliability-constrained environments.
The annual base salary range for this San Francisco Bay Area position is $200,000–$260,000. Health coverage, benefits, and generous equity come with the role. This role is on-site in the San Francisco Bay Area. A relocation bonus is available. We hire to start as soon as possible. We review what you’ve shipped; if it’s the caliber we seek, interviews take about a week, start to finish. We treat your work and conversations with discretion.
