Sancho Robotics builds general-purpose mobile manipulators for advanced manufacturing. Our work was featured in the opening of the NVIDIA GTC 2026 keynote. Backed by $7 million in funding, our team has won multiple best-paper awards at top robotics conferences such as RSS and IROS and brings a decade of real-world robotics deployment experience.

As a core founding member, you will define and build the autonomy stack that allows our robots to understand and operate in dynamic 3D environments. Your initial focus will be perception: turning modern perception models and robot data into reliable capabilities on physical robots.

What you will do

• Implement autonomy modules: Implement reliable, production-quality components that give our robots a coherent understanding of their 3D environment.

• Debug cross-module failures: Use diagnostic tools like visualization, logs, and real robot data to diagnose failures of a complex robotics system.

• Manage system tradeoffs: Make practical engineering decisions across accuracy, latency, compute, robustness, and real-time robot behavior.

• Test on real robots: Use hardware experiments and field data to evaluate performance, identify failure modes, and improve the autonomy reliability.

What we are looking for

• Strong C++ and Python skills: you write robust and production-quality robotics software.

• You have deployed and tested on physical robots and sensors in the real world, not just simulation.

• Solid robotics fundamentals with hands-on experience in perception, planning, or manipulation.

• You understand how autonomy modules interact in a real robot system and can reason about tradeoffs across accuracy, latency, robustness, and compute.

• Familiar with ROS2 or similar middleware.

Bonus points

• Deep understanding of modern foundation perception models: DINO, DETR, Segment Anything, VLMs, or related architectures.

• Experience building low-latency inference systems with CUDA or Triton on edge devices such as NVIDIA Orin.