Applied Scientist - Agents & Robotics
- Onsite
- Full Time
About the role
We are looking for highly motivated Master's and PhD graduates to join our team full-time in an invigorating environment where researchers and engineers are building the future of robotics and learning-based systems. Strong candidates will bring exceptional talent with solid fundamentals in agents and system design. In this role, you will focus on developing and scaling agentic infrastructure and systems for robot autonomy.
If you have a passion for robots and a desire to be part of the advancement of robot capabilities and usefulness, you will want to join us!
What you'll do
Build, extend, and maintain an agentic framework for managing on-robot workflows across various real-world deployment scenarios.
Create and manage internal benchmarks for the comprehensive evaluation of agentic workflows in diverse robotic tasks.
Work closely with engineers and researchers to integrate your work with powerful simulation tools and on-robot deployments.
Profile, benchmark, and eliminate performance bottlenecks in resource-constrained Linux environments to ensure production-grade reliability.
What you'll need
A completed Master's (MS) or PhD degree in Robotics, Computer Science, Computational Science & Engineering, or a related field.
Proven software engineering experience (e.g., industry internships, significant open-source contributions, or complex academic/industry project work).
AI/ML Foundations: Experience with PyTorch and a strong understanding of modern AI architectures (LLMs, VLMs, or Reinforcement Learning).
Bonus
While strong core software engineering fundamentals in systems programming are required, no specific domain expertise beyond that is necessary. However, if you happen to possess experience in any single one of the following areas, it would be a massive plus and highly relevant to our current projects:
Sim-to-Real Experience: Experience in deploying ML workflows on physical hardware.
Vision-Language Models (VLMs): Experience integrating multimodal models for robotic perception and task planning.
Simulators: Experience testing code in simulated environments (e.g., Isaac Sim, MuJoCo).
Robotic Middleware: Familiarity with ROS2.
Robotic Perception: Familiarity with various scene representations (e.g., Gaussian Splattings, Scene Graphs, World Models, SDFs, etc.), perception frontends (e.g., DINO, SAM, SiGLIP, etc.), and approaches to scene understanding.
Robot Learning: Experience in training ML models for robot action policies, robot perception frontends, and state estimation.