I’m an experienced generalist and systems thinker with the proven ability to deliver solutions for challenging practical applications. I have the curiosity to find or develop the best tools for the job, a drive to continuous improvement, and the perseverance to see things through from concept to production. I am a strong individual contributor, and I have management experience.

Key skills and interests: Machine Learning (Deep Neural Networks, Reinforcement Learning); Robotics (Motion Planning, Geometry, Simulation); Visualization; Software Design and Software Development Methodology.

References are available on request.

Selected Experience

Staff Research Scientist — Intrinsic (Alphabet); Mountain View, California, USA – 2022-2023

“Make industrial robotics accessible and usable for millions more businesses, entrepreneurs, and developers.”

I’m in Stefan Schaal’s Research and Technology Transfer group. Our role is to bring state of the art methods to Intrinsic’s platform and build tooling for research and development. I am fully remote.

Topics: Physics Simulation (Bullet, Isaac Sim/Gym); Reinforcement Learning for Manipulation; Neural Implicit Representations (DeepSDF, NeRF).

Staff Research Scientist / Team Lead — Vicarious; Union City, California, USA – 2017-2022

“Make fundamental advances in AI, machine learning, and computer vision to build a flexible robotics platform.”

Vicarious was acquired by Intrinsic in May 2022.

I held various positions at Vicarious throughout my tenure and made wide ranging contributions. I worked fully remotely from 2020 onwards.

Staff Research Scientist (2021-2022). Vicarious’ original software was written for single robots, but increasingly our deployed systems were made up of many robots, conveyor belts and other equipment. I developed a prototype system level representation for coordination and planning. I also did R&D on (Neural) Implicit Geometry representations (e.g. for faster collision checking, object completion, data generation).

Simulation Lead (2019-2021). I proposed and started this team. We deployed a simulation stack (Ignition Gazebo) and we made it run reliably in Continuous Integration (CI) Testing (not trivial). Being able to test most software changes in simulation was crucial for scalability.

Motion Planning Lead (2018-2019). We were responsible for developing and maintaining fast and robust motion planners as well as inverse kinematics for production systems. Runtime was key to reducing cycle time and determinism was key for both development and user experience. We developed various supporting benchmarking and visualization tools. Before being promoted to lead, I was a member of the team.

Key out-of-role contributions. Designed and implemented an Entity Component System (ECS) world model and visualization tools. Proposed and led conversion of our world model format to unlock major improvements for scalability, built tooling for supporting the new format. Developed the company standard for maintainable documentation.

Scientist / Intern — Brain Corporation; San Diego, California, USA – 2014-2017

“Create software that transforms everyday machines into autonomous solutions.”

Scientist (2015-2017). Proposed and developed a mapping solution (SLAM) for large indoor environments (e.g. Walmart superstores), and made it work in challenging production environments. Started solo but eventually became tech lead for a three person team. To the best of my knowledge, this solution is still the core of what is in use today, deployed on 15,000+ robots across the world. Multiple patents awarded for the algorithm, how to make it work in the real world, and robot UX. I’m the main inventor on two.

Intern (2014-2015). There were various competing ideas for Brain Corp’s core product and I was part of the team that proposed the solution that the company pivoted to (floor cleaning robots). I helped design the prototype and pushed for key technical decisions.

Selected Education

PhD — Center for Math and Computer Science / VU / Netherlands Institute for Neuroscience; Amsterdam, NL – 2009-2015

Supervisors: Sander Bohte, Pieter Roelfsema

Developed a biologically plausible learning rule to train neural networks equipped with working memory using reinforcement learning. My key publications were at NeurIPS and in PLoS Computational Biology.
I won an excellent reviewer award from the NeurIPS program committee.

Fun fact: The Python programming language originates at CWI.

MSc: Artificial Intelligence — University of Groningen, NL / Carnegie Mellon University (CMU), USA – 2006-2009

Cum laude. I specialized in Machine Learning and Robotics. I did my thesis project at Carnegie Mellon’s Robotics Institute on “Detection and recognition of persons with a static robot”.

BSc: Artificial Intelligence — University of Groningen, NL – 2003-2006

With honors. Thesis project: “Co-evolution and speciation”.