1.1 SUMI Lab

1.1 SUMI Lab

1.1 SUMI Lab

We're the Sequential Uncertainty Monitoring and Interpretability lab at TU Delft, led by Anna Lukina. We design mathematical formalisms and efficient algorithms to explore new ways to verify and monitor AI systems.

Our main goal is to build fast, verified, and interpretable components for safety-critical systems, at a reasonable cost. We think we can get there with a combination of neural training, small decision trees, control theory, stochastic mathematics, efficient abstraction, constraint-based optimization, and search. We host a regular SUMI Summit to explore state-of-the-art papers and brainstorm with our international collaborators.

Learn more about our research or explore our student project opportunities. Beyond research, we care about teaching, in particular about how to democratize AI verification.

At TU Delft, we are involved in teaching bachelor's and master's courses. Explore our course material

Our main goal is to build fast, verified, and interpretable components for safety-critical systems, at a reasonable cost. We think we can get there with a combination of neural training, small decision trees, control theory, stochastic mathematics, efficient abstraction, constraint-based optimization, and search. We host a regular SUMI Summit to explore state-of-the-art papers and brainstorm with our international collaborators.

Learn more about our research or explore our student project opportunities. Beyond research, we care about teaching, in particular about how to democratize AI verification.

At TU Delft, we are involved in teaching bachelor's and master's courses. Explore our course material

Our main goal is to build fast, verified, and interpretable components for safety-critical systems, at a reasonable cost. We think we can get there with a combination of neural training, small decision trees, control theory, stochastic mathematics, efficient abstraction, constraint-based optimization, and search. We host a regular SUMI Summit to explore state-of-the-art papers and brainstorm with our international collaborators.

Learn more about our research or explore our student project opportunities. Beyond research, we care about teaching, in particular about how to democratize AI verification.

At TU Delft, we are involved in teaching bachelor's and master's courses. Explore our course material

We are excited to explore new collaborations, and the lab accepts master’s and bachelor’s students every year. If you are interested in AI verification, probabilistic reasoning, and optimization algorithms, get in touch!

SUMI lab was created in October 2023 and will soon move to its own webpage.

1.2 SUMI Team

1.2 SUMI Team

1.2 SUMI Team

Meet the team:

Anna Lukina

Team Leader

Leading research, education, and valorization in AI Verification at TU Delft and internationally.

Anna Lukina

Team Leader

Leading research, education, and valorization in AI Verification at TU Delft and internationally.

Anna Lukina

Team Leader

Leading research, education, and valorization in AI Verification at TU Delft and internationally.

Sterre Lutz

PhD Researcher

Developing theory and algorithms for safe, interpretable and high-performing control of AI systems.

Sterre Lutz

PhD Researcher

Developing theory and algorithms for safe, interpretable and high-performing control of AI systems.

Sterre Lutz

PhD Researcher

Developing theory and algorithms for safe, interpretable and high-performing control of AI systems.

Daniël Vos

Postdoctoral Researcher

Designing algorithms for interpretable sequential decision making.

Daniël Vos

Postdoctoral Researcher

Designing algorithms for interpretable sequential decision making.

Daniël Vos

Postdoctoral Researcher

Designing algorithms for interpretable sequential decision making.

Jeff Smits

Research Engineer

Developing an interactive platform for decision-tree controlled systems.

Jeff Smits

Research Engineer

Developing an interactive platform for decision-tree controlled systems.

Jeff Smits

Research Engineer

Developing an interactive platform for decision-tree controlled systems.

1.3 Open projects

1.3 Open projects

1.3 Open projects

If you're interested in being part of my group, check out open projects!

I'm generally interested in the intersection of formal methods and machine learning (especially for control and verification), as well as topics around interpretability and visualizations.


There's a brief summary of my background on my CV page, and more recent information on my lab's research page.


  • If you are looking for projects and research directions that I would be excited to supervise or collaborate on right now, check out the list.

  • If you want to know more about my medium and long-term goals / perspectives, check out the research summary (pdf).

  • If you care about teaching, or if you're interested in an academic career in the long term, you may also want to have a look at my statement on teaching (pdf).

  • If you have concrete questions about SUMI Lab, send me an email.

1.4 Outgoing research visits

1.4 Outgoing research visits

1.4 Outgoing research visits

🇯🇵

January 5 – February 13, 2026

Prof. Ryo Kuroiwa (黒岩稜), National Institute of Informatics, Tokyo, Japan.

🇨🇭

October 27–28, 2025

Prof. Clément Pit-Claudel, EPFL, Switzerland.

🇯🇵

October 28 – November 1, 2024

Prof. Masaki Waga (和賀 正樹), Kyoto University, Japan.

🇩🇪

September 12–13, 2024

Prof. Bernd Finkbeiner, CISPA, Germany.

🇺🇸

July 1–31, 2024

🇦🇹

May 5–7, 2022

Univ.-Prof. Mag.art Manuela Naveau PhD, Interface Cultures at The University of Art and Design Linz, and Nicolas Naveau, FUTURELAB at Ars Electronica, Linz, Austria.

🇩🇰

April 15–22, 2022

🇺🇸

January 11 – May 31, 2021

Simons Institute for the Theory of Computing, Berkeley, USA. Research Fellow under mentorship of Dr. Pavithra Prabhakar.

🇦🇺

April 1, 2020

Prof. James Bailey, Prof. Peter J. Stuckey, Dr. Emir Demirović, University of Melbourne, Melbourne, Australia.

🇦🇺

February 11 – March 12, 2019

Prof. James Bailey, Prof. Peter J. Stuckey, Dr. Emir Demirović, University of Melbourne, Melbourne, Australia.

🇯🇵

February 1 – May 31, 2018

Prof. Fuyuki Ishikawa, Ishikawa Lab, National Institute of Informatics, and Prof. Ichiro Hasuo, ERATO MMSD, Tokyo, Japan.

🇩🇪

September 3 – December 31, 2017

Prof. Joost-Pieter Katoen, MOVES Group, RWTH Aachen University, Germany.

🇺🇸

May 15 – July 31, 2017

Prof. George Pappas and Prof. Vijay Kumar, GRASP Lab at PERCH, University of Pennsylvania, Philadelphia, PA, USA.

1.5 Incoming research visits

1.5 Incoming research visits

1.5 Incoming research visits

🇩🇰

Prof. Christian Schilling, Aalborg University, Denmark.

🇯🇵

Prof. Masaki Waga (和賀 正樹), Kyoto University, Japan.

🇩🇰

Prof. Christian Schilling, Aalborg University, Denmark.

🇦🇹

Dr. Andre Schidler, TU Wien, Austria.

🇦🇹

Prof. Djordje Zikelic (Đorđe Žikelić), Singapore Management University.

🇬🇧

Prof. Mirco Giacobbe, University of Birmingham, UK.

🇩🇰

Prof. Christian Schilling, Aalborg University, Denmark.

🇺🇸

Dr. Florent Delgrange, Vrije Universiteit Brussel (VUB), Belgium.

1.6 SUMI Alumni

1.6 SUMI Alumni

1.6 SUMI Alumni

Aaron Berger

PhD Researcher

Applied formal methods to reliable automation in the railways industry.

Aaron Berger

PhD Researcher

Applied formal methods to reliable automation in the railways industry.

Johannes Koch

PhD Researcher

Improved genetic programming for explainable AI in collaboration with CWI and LUMC.

Johannes Koch

PhD Researcher

Improved genetic programming for explainable AI in collaboration with CWI and LUMC.

Nathalie van de Werken

MSc, 2025

"Reducing Uninteresting Anomalies".

Nathalie van de Werken

MSc, 2025

"Reducing Uninteresting Anomalies".

Daan Hofman

MSc, 2023

"VoBERT: Unstable Log Sequence Anomaly Detection" with Yury Zhauniarovich and Eduardo Barbaro from ING.

Daan Hofman

MSc, 2023

"VoBERT: Unstable Log Sequence Anomaly Detection" with Yury Zhauniarovich and Eduardo Barbaro from ING.

Daniël van Gelder

MSc, 2022

"Real-Time Passenger Load Estimation using In-Vehicle Data" with Siemens and Oded Cats.

Daniël van Gelder

MSc, 2022

"Real-Time Passenger Load Estimation using In-Vehicle Data" with Siemens and Oded Cats.

Erik Sennema

MSc, 2023

"Elastic gradient boosting decision trees under limited labels by sequential epistemic uncertainty quantification: Elastic CatBoost Uncertainty (eCBU)" with Yury Zhauniarovich and Eduardo Barbaro from ING.

Erik Sennema

MSc, 2023

"Elastic gradient boosting decision trees under limited labels by sequential epistemic uncertainty quantification: Elastic CatBoost Uncertainty (eCBU)" with Yury Zhauniarovich and Eduardo Barbaro from ING.

Rens Oude Elferink

MSc, 2024

"Identifying how drivers adapt to automated vehicles by monitoring neural networks" with Luciano Cavalcante Siebert.

Rens Oude Elferink

MSc, 2024

"Identifying how drivers adapt to automated vehicles by monitoring neural networks" with Luciano Cavalcante Siebert.

Elwin Duinkerken

MSc, 2023

"Robust Shunting in a Dynamic Environment: Deriving Proactive Schedules from a Reactive Policy" with the Dutch Railways.

Elwin Duinkerken

MSc, 2023

"Robust Shunting in a Dynamic Environment: Deriving Proactive Schedules from a Reactive Policy" with the Dutch Railways.

Cas van Rijn

MSc, 2023

"Combining Multi-Objective Planning with Reinforcement Learning to Solve Complex Tasks in Environments with Sparse Rewards."

Cas van Rijn

MSc, 2023

"Combining Multi-Objective Planning with Reinforcement Learning to Solve Complex Tasks in Environments with Sparse Rewards."

Parand Alizadeh Alamdari

MSc, 2020

"Formal Methods with a Touch of Magic."

Parand Alizadeh Alamdari

MSc, 2020

"Formal Methods with a Touch of Magic."