David Leeftink

Radboud University

Nijmegen, the Netherlands

david.leeftink (at) ru.nl.

I am a PhD candidate at the Donders Institute for Brain, Cognition and Behaviour (Radboud University), supervised by Dr. Max Hinne and Prof. Marcel van Gerven.

My research focuses on reinforcement learning and control for dynamical systems under uncertainty. I approach deep RL through the lens of optimal control theory, aiming to understand and improve decision-making under uncertainty in dynamical systems. My goal is to develop methods that are robust and deployable in real-world systems.

A central question in my work is: how can we make reliable decisions under model uncertainty? To address this, my work centers on three directions:

Probabilistic control under uncertainty: optimal control methods for systems with unknown or partially known dynamics.
Data-efficient optimization of physical systems: Bayesian optimization for high-cost and real-world settings, such as semiconductor manufacturing and neural implants.
Deep Reinforcement Learning: model-based and model-free reinforcement learning from a dynamical systems perspective, aiming to improve data-efficiency and interpretability.

Ultimately, my goal is to root modern reinforcement learning algorithms in the notions of robustness, safety, and stability.

news

Mar 15, 2026	Paper accepted: Our recent work on Bayesian Optimization for Semiconductor Manufacturing has been accepted for IFAC’s Control Engineering Practice.
Dec 09, 2025	Upcoming Talk: I presented our recent work on mean hamiltonian minimization in a lightning round talk at the Workshop on Stochastic Planning & Control of Dynamical Systems at CDC in Rio de Janeiro, on December 9th 2025.
Jul 17, 2025	Paper Acceptance: Our work on Mean Hamiltonian Minimization has been accepted at IEEE Conference for Decision and Control (CDC) 2025.
Jun 12, 2025	Upcoming Talk: I gave a contributed talk at the Workshop on Theory of Control and Reinforcement Learning at CWI, Amsterdam on the 19th of June 2025.

latest posts

selected publications

IEEE CDC

Optimal Control of Probabilistic Dynamics Models via Mean Hamiltonian Minimization

David Leeftink, Çağatay Yıldız, Steffen Ridderbusch, Max Hinne, and Marcel Van Gerven

In 2025 IEEE 64th Conference on Decision and Control (CDC), 2025

DOI Bib HTML PDF Code

@inproceedings{leeftink2025optimalcontrolprobabilisticdynamics,
  author = {Leeftink, David and Yıldız, {\c C}a{\u g}atay and Ridderbusch, Steffen and Hinne, Max and Van Gerven, Marcel},
  booktitle = {2025 IEEE 64th Conference on Decision and Control (CDC)},
  doi = {10.1109/CDC57313.2025.11312001},
  keywords = {Uncertain systems;Uncertainty;System dynamics;Optimal control;Reinforcement learning;Ordinary differential equations;Probabilistic logic;Minimization;Mathematical models;Numerical models},
  pages = {4146-4153},
  title = {Optimal Control of Probabilistic Dynamics Models via Mean Hamiltonian Minimization},
  year = {2025},
  bdsk-url-1 = {https://doi.org/10.1109/CDC57313.2025.11312001}
}

IFAC CEP
Automated discovery of laser dicing processes with Bayesian optimization for semiconductor manufacturing

David Leeftink, Roman Doll, Heleen Visserman, Marco Post, Faysal Boughorbel, and 2 more authors

IFAC Control Engineering Practice, 2026

Abs DOI Bib HTML

Laser dicing of semiconductor wafers is a critical step in microelectronic manufacturing, where multiple sequential laser passes precisely separate individual dies from the wafer. Adapting this complex sequential process to new wafer materials typically requires weeks of expert effort to balance process speed, separation quality, and material integrity. We present the first automated discovery of production-ready laser dicing processes on an industrial LASER1205 dicing system. We formulate the problem as a non-linear, constrained multi-objective Bayesian optimization task, and introduce a sequential two-level fidelity strategy to minimize expensive destructive die-strength evaluations. On bare silicon and product wafers, our method autonomously delivers feasible configurations that match or exceed expert baselines in production speed, die strength, and structural integrity, using only technician-level operation. Post-hoc validation of different weight configurations of the utility functions reveals that multiple feasible solutions with qualitatively different trade-offs can be obtained from the final surrogate model. Expert-refinement of the discovered process can further improve production speed while preserving die strength and structural integrity, surpassing purely manual or automated methods.
@article{leeftink2026laserdicing, title = {Automated discovery of laser dicing processes with Bayesian optimization for semiconductor manufacturing}, journal = {IFAC Control Engineering Practice}, volume = {172}, pages = {106941}, year = {2026}, issn = {0967-0661}, doi = {https://doi.org/10.1016/j.conengprac.2026.106941}, author = {Leeftink, David and Doll, Roman and Visserman, Heleen and Post, Marco and Boughorbel, Faysal and Hinne, Max and {van} Gerven, Marcel}, keywords = {Bayesian optimization, Semiconductor manufacturing, Laser dicing, Laser separation, Machine learning, Process optimization}, }