Summary
Pavel Czempin is a research-focused software engineer and PhD student at USC with 14 years of experience blending machine learning research and engineering. He has deep expertise in reinforcement learning, RLHF, and adversarial defenses from sustained collaborations with Berkeley AI Research and hands-on internships that produced a first-author ICML workshop paper. Pavel builds reproducible ML systems in Python and PyTorch, with applied work on audio deepfake detection and annotator-expertise estimation for RL from human feedback. He combines academic teaching experience in functional programming and verification with practical software development dating back to embedded and Java tooling projects. Based in Los Angeles, he brings a rare mix of rigorous research, production-minded implementation, and cross-cultural academic training from TUM and NTU. Notably, his trajectory shows a consistent pivot from classical engineering tools toward cutting-edge safety and interpretability in RL.
13 years of coding experience
2 years of employment as a software developer
Computer Science, Computer Science at Nanyang Technological University
Nanodegree, Deep Learning, Nanodegree, Deep Learning at Udacity
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Southern California
Bachelor's degree, Engineering Science, Bachelor's degree, Engineering Science at Technical University Munich
Master of Science - MS, Computer Science, Master of Science - MS, Computer Science at Technical University of Munich