Summary
Daniel Pickem is a Staff Research Engineer specializing in reinforcement learning and post-training algorithms, currently driving Nemotron model development at NVIDIA after 11 years in industry and academia. He blends deep research credentials—a PhD in Robotics from Georgia Tech and postdoctoral leadership of the Robotarium swarm-robotics testbed—with hands-on production engineering at Apple and NVIDIA, shipping large-scale evaluation and distributed data pipelines for autonomy. At NVIDIA he has bridged foundation models into evaluation workflows (LLM-as-judge) and now focuses on scaling NeMo-RL, Megatron Core, and the NeMo Framework for post-training at production scale. His background spans robotics hardware design, perception and control, GPU-accelerated numerical kernels, and practical ML systemization, giving him rare end-to-end fluency from low-level performance tuning to ML product metrics and dashboards. Based in California, he combines academic rigor with product-focused execution and a track record of turning complex research prototypes into robust, widely used infrastructure. An uncommonly useful facet of his profile is the mix of swarm-robotics testbed leadership and foundation-model engineering, enabling both physical-system intuition and large-model deployment expertise.
11 years of coding experience
9 years of employment as a software developer
B.Sc., Electrical Engineering and Computer Science, B.Sc., Electrical Engineering and Computer Science at Vienna University of Technology
Doctor of Philosophy (Ph.D.), Robotics, Doctor of Philosophy (Ph.D.), Robotics at Georgia Institute of Technology
German, English, Russian, French