Research & Developement in Embodied AI and Robotics

R&D at a Stealth-Mode Robotics Startup

At a stealth-mode robotics AI startup, my work focuses on the full lifecycle of research and development for novel embodied AI systems.

I am responsible for both building foundational R&D infrastructure and designing and deploying core algorithms for high-precision autonomy. This work enables our mobile robotics platforms to operate effectively despite severe hardware and sensor constraints.

Below are two key projects detailing this experience. As this work is for a startup in stealth, I am limited in the specific details I can share. All descriptions focus on the technical problems and my solutions while protecting proprietary information.

Project 1: Robust Robotic State Estimation Algorithm

A novel, on-device AI system I researched, designed, and deployed from first principles. This algorithm uses a manifold disentanglement technique to overcome the failures of SOTA methods on our noisy, resource-constrained sensors. The system is implemented in Lisp and achieved a >100x improvement in state estimation accuracy.

Project 2: 3D Robotic Simulation & Analysis Platform

A high-fidelity simulation and analysis tool I built from the ground up using Unreal Engine. It serves as a testbed available company-wide for validating new autonomous features, analyzing system performance, and debugging complex, real-world mission data.