Lab Visits 2026.1

23 Mar 2026, 09:00 → 24 Mar 2026, 17:00 Europe/Berlin

Lamarr/RC Trust Dortmund

Jens Buß (Lamarr Institute, TU Dortmund University), Mirko Bunse (Lamarr Institute, TU Dortmund University)

Welcome to the Lamarr Lab Visits! This event provides a setting to network, exchange ideas and collaborate beyond the physical and disciplinary borders of the Lamarr locations and research areas.

In this edition of the event series, TU Dortmund University provides the meeting rooms. The program will be filled with sessions and activities that are proposed by the attendants. Topics include specific research questions to be discussed, research area meetings, informal exchanges about novel scientific developments, and hard-skill training sessions. The program rounds off with a social dinner.

Please refer to this event page for all information on the event. We look forward to seeing many of you very soon!

Contact

If you encounter any issues or questions, please reach out to Mirko and Jens, the local organizers.

Monday 23 March

Arrival

Location

Joint Programm: Kick-Off Session

Location

1 Welcome & Opening

Speaker: Prof. Jakob Rehof (TU Dortmund)

2 Area Pitches

First part of the area updates with pitches of 5min per area presenting updates from the areas and a summary of current research topics.
Involved Areas:

• Resource-Aware ML
• Embodied AI
• Planning & Logistics
• Industry & Production
• Life Sciences & Health

The remaining areas will be presented on day 2.

Speaker: Nilusha Aliman

11:45 Break

3 Science Speed Dating

Speaker: Dr Jens Buß (Lamarr Institute, TU Dortmund University)

4 Session Pitches

Introduction of the session offered during the first day (2 min/Session)

Speaker: Nilusha Aliman

5 Orga

Speakers: Dr Jens Buß (Lamarr Institute, TU Dortmund University), Nilusha Aliman

Orga.pptx

12:55 Group Foto

13:00 Lunch (with some extra time for handling luggage)

(subsequent program takes place at Joseph-von-Fraunhofer-Straße 25, 44227 Dortmund)

LogSense: Benchmarking of Domain-Specific LLMs & LLM-assisted Dataset Creation

Conveners: Christian Pionzewski (Fraunhofer IML), Christina Braun

6 LogSense: Benchmarking of Domain-Specific Language Models & Agentic Dataset Creation

Speakers: Christian Pionzewski (Fraunhofer IML), Christina Braun

RAML: Benchmarking and Trade-Offs in AI & Efficiency Tracking with LET

Conveners: Raphael Fischer (TU Dortmund University), Sebastian Buschjäger (Lamarr Institute for ML and AI, TU Dortmund)

7 RAML: Benchmarking and Trade-Offs in AI & Efficiency Tracking with LET

Speakers: Raphael Fischer (TU Dortmund University), Sebastian Buschjäger (Lamarr Institute for ML and AI, TU Dortmund)

Towards an Era of Posthuman Logistics

Conveners: Christoph Pott, Julian Eßer, Moritz Roidl

8 Towards an Era of Posthuman Logistics

Speakers: Christoph Pott, Julian Eßer, Moritz Roidl

Error analyses for deep learning methods for partial differential equations

Conveners: Arnulf Jentzen, Benno Kuckuck, Dr Ramsés Sánchez (Lamarr institute, University of Bonn)

9 Error analyses for deep learning methods for partial differential equations

Speakers: Arnulf Jentzen, Benno Kuckuck, Dr Ramsés Sánchez (Lamarr institute, University of Bonn)

Physics LLM

Convener: Tim Ruhe (TU Dortmund)

10 Physics LLM

Speaker: Tim Ruhe (TU Dortmund)

Workshop: Batteries included - Open Source CLI Coding Assistants and Agents

This session will explore several powerful open-source CLI coding assistants in a hands-on 90-minute workshop. Learn or share your experience on how to integrate AI-assisted development into your research workflow while maintaining full control through open-source tools and custom LLM endpoints. Experiment with different agents, compare their strengths, and start using them immediately in your projects.

Register now

Space is limited to 8 participants, first come, first serve.

---

The landscape of AI-assisted coding has evolved dramatically, with exploding options of promises of powerful programming partners.
Open-source alternatives combined with institution-provided models offer transparency, flexibility, and cost-effectiveness. This workshop introduces some emerging, mature or production-ready (?) CLI coding assistants designed for coders who value flexibility and like terminal-based workflows.
Rather than passive lectures, this session emphasizes practical experimentation. You should bring your own code for exploration (or at least some ideas what to tackle from scratch).
By workshop's end, participants will understand how to integrate these open-source agents into their daily research workflows, customize them for specific languages and frameworks, and maintain full transparency over model interactions through open-source infrastructure.

---

Prerequisites:

• Your own laptop, obviously, with working eduroam Wifi, as we will need the network a lot.
• Operating system: Linux, Mac OS or Windows with WSL2 recommended, pure Windows depends on the specific assistant.
• You should be comfortable working in a terminal/command line. This workshop uses terminal-based tools exclusively to stay independent of specific IDEs.
• Optional: Container technology like Docker or Podman (recommended) preinstalled if you would like to encapsulate the assistants and their installation from your work environment and for added safety.
• Optional: Bring your own code.

Convener: Stefan Michaelis (Lamarr Institute - TU Do)

11 Workshop: Batteries included - Open Source CLI Coding Assistants and Agents

This session will explore several powerful open-source CLI coding assistants in a hands-on 90-minute workshop. Learn or share your experience on how to integrate AI-assisted development into your research workflow while maintaining full control through open-source tools and custom LLM endpoints. Experiment with different agents, compare their strengths, and start using them immediately in your projects.

Register now

Space is limited to 8 participants, first come, first serve.

---

The landscape of AI-assisted coding has evolved dramatically, with exploding options of promises of powerful programming partners.
Open-source alternatives combined with institution-provided models offer transparency, flexibility, and cost-effectiveness. This workshop introduces some emerging, mature or production-ready (?) CLI coding assistants designed for coders who value flexibility and like terminal-based workflows.
Rather than passive lectures, this session emphasizes practical experimentation. You should bring your own code for exploration (or at least some ideas what to tackle from scratch).
By workshop's end, participants will understand how to integrate these open-source agents into their daily research workflows, customize them for specific languages and frameworks, and maintain full transparency over model interactions through open-source infrastructure.

---

Prerequisites:

• Your own laptop, obviously, with working eduroam Wifi, as we will need the network a lot.
• Operating system: Linux, Mac OS or Windows with WSL2 recommended, pure Windows depends on the specific assistant.
• You should be comfortable working in a terminal/command line. This workshop uses terminal-based tools exclusively to stay independent of specific IDEs.
• Optional: Container technology like Docker or Podman (recommended) preinstalled if you would like to encapsulate the assistants and their installation from your work environment and for added safety.
• Optional: Bring your own code.

Speaker: Stefan Michaelis (Lamarr Institute - TU Do)

Register for Workshop

slides.pdf

Exchange on Interpretability

Conveners: Katharina Beckh, Lovis Heindrich

12 Exchange on Interpretability

Speakers: Katharina Beckh, Lovis Heindrich

How to raise third-party funds: Learning from one another in Lamarr.

Acquiring third-party funding is an imperative for any researchers keen on pursuing a scientific career. The session is aimed at (early career) Lamarr members who are planning to apply for third-party funds/grants in the foreseeable future. The goal is to provide them with a framework in which they can learn from experienced Lamarr members, get tips, and ask questions—at the same time, the mutual exchange could also provide inspiration for experienced researchers.
Johannes Albrecht, Zorah Lähner and Arnulf Jentzen have kindly agreed to contribute their expertise and answer your questions!

Conveners: Ann-Kathrin Oster (Lamarr Institute at TU Dortmund University), Arnulf Jentzen, Johannes Albrecht (TU Dortmund & LAMARR), Zorah Lähner (University of Bonn)

13 How to raise third-party funds: Learning from one another in Lamarr.

Acquiring third-party funding is an imperative for any researchers keen on pursuing a scientific career. The session is aimed at (early career) Lamarr members who are planning to apply for third-party funds/grants in the foreseeable future. The goal is to provide them with a framework in which they can learn from experienced Lamarr members, get tips, and ask questions—at the same time, the mutual exchange could also provide inspiration for experienced researchers.
Johannes Albrecht, Zorah Lähner and Arnulf Jentzen have kindly agreed to contribute their expertise and answer your questions!

Speakers: Ann-Kathrin Oster (Lamarr Institute at TU Dortmund University), Arnulf Jentzen, Johannes Albrecht (TU Dortmund & LAMARR), Zorah Lähner (University of Bonn)

17:30 Time for Check-In

Time for transfer to your Hotel and check-In there

14 Social Dinner @ Restaurant "Mit Schmackes"

On Monday evening at 7:00 pm, we’ll head together to Mit Schmackes for our Social Dinner. Expect hearty Ruhrpott-style comfort food, bold flavors, and a relaxed, down-to-earth atmosphere where everyone feels welcome. It’s a great place for conversation and exchange in a cozy, informal setting.

Tuesday 24 March

Joint Programm: Morning Kick-Off Session

Location

15 Area Pitches

Second part of the area updates with pitches of 5min per area presenting updates from the areas and a summary of current research topics.
Involved Areas:

• Hybrid ML
• Trustworthy AI
• Physics
• Natural Language Processing

Speaker: Nilusha Aliman

16 Session Pichtes

Speaker: Dr Jens Buß (Lamarr Institute, TU Dortmund University)

Scientifc Coordinator Meeting

Meeting of the Scientific Coordinators (Closed Session)

Convener: Dr Jens Buß (Lamarr Institute, TU Dortmund University)

Transfer Workshop: Research to Use - Exploring Ways to Create Impact with Your Research

This session creates a space for exchange and collaborative discussion on how research ideas can generate value beyond academia. Together, we will explore pathways for transforming research findings into inventions and identify opportunities for knowledge and technology transfer — areas that are becoming increasingly important in today’s scientific, societal, and economic landscape.
We will address questions such as:
How can your research create impact beyond the academic sphere?
What pathways to impact exist, and how can Lamarr support you along the way?

Conveners: Andreas Becker (CET), Daniela Treutlein (enaCom), Heike Horstmann (Fraunhofer IAIS), Dr Nilusha Aliman

Workshop Registration

17 Transfer Workshop: Research to Use - Exploring Ways to Create Impact with Your Research

This session creates a space for exchange and collaborative discussion on how research ideas can generate value beyond academia. Together, we will explore pathways for transforming research findings into inventions and identify opportunities for knowledge and technology transfer — areas that are becoming increasingly important in today’s scientific, societal, and economic landscape.
We will address questions such as:
How can your research create impact beyond the academic sphere?
What pathways to impact exist, and how can Lamarr support you along the way?

register here

Lamarr Collection on Trustworthy AI: What is trustworthiness in Lamarr's scientific areas, and why does it matter?

Conveners: Eva Schmidt, Prof. Jakob Rehof (TU Dortmund), Sara Mann

18 Lamarr Collection on Trustworthy AI

What is trustworthiness in Lamarr's scientific areas, and why does it matter?

Speakers: Eva Schmidt, Prof. Jakob Rehof (TU Dortmund), Sara Mann

Recording Data for Robotic Foundation Models

Convener: Hermann Blum

19 Recording Data for Robotic Foundation Models

Speaker: Hermann Blum

13:00 Lunch

1st Meeting of the Amortized Inference (Reading) Group

Conveners: Patrick Seifner (University of Bonn), Dr Ramsés Sánchez (Lamarr institute, University of Bonn)

Quantum Algorithm Discovery

Convener: Lukas Theißinger

NLP for Science

Convener: Florian Mai (University of Bonn)

20 NLP for Science

Speaker: Florian Mai (University of Bonn)

Next Generation of AI Agents

Conveners: Héctor Allende-Cid, Najmeh Mousavi

Physics: Lightning Talks & Discussion -- What We’re Working On

This session offers a fast-paced overview of ongoing physics research within the Lamarr Institute, featuring concise 15-minute talks on large collaborations and funded projects. Alongside these, early-stage ideas and subprojects—such as PhD and Master’s thesis topics—are pitched to spark discussion and potential collaboration. The format is designed to highlight both established work and emerging opportunities in a dynamic, engaging way.

Convener: Dr Jens Buß (Lamarr Institute, TU Dortmund University)

21 Introduction: Wheel of Physics

Speaker: Dr Jens Buß (Lamarr Institute, TU Dortmund University)

22 NuRadioOpt

Speaker: Christian Glaser (TU Dortmund University)

23 NuRadioOpt: First Results of Differentiable Optimization

Please refer to this contribution for details.

Speaker: Nicolai Weitkemper (TU Dortmund University)

24 NuRadioOpt: Neural Network Trigger

Speakers: René Reimann, Yannik Schaper (TU Dortmund, Physik, Astroparticle Physics)

25 Einstein Telescope

Speaker: Dominik Elsässer

26 TBD

Speaker: Luca Di Bella

27 Reconstruction of Radio Interferometer Measurements of Protoplanetary Disks using radionets

This pitch outlines the efforts to reconstruct the missing information in radio interferometer measurements of protoplanetary disks using the deep-learning framework radionets.
For that we are currently generating training data using hydrodynamics models and MC simulations to enhance the scientific interpretation of the reconstructions.

Speaker: Tom Groß (TU Dortmund University)

presentation.pdf

28 BRAID

Speaker: Christian Glaser (TU Dortmund University)

29 Construction of a 2D Neutrino Sky Map with IceCube Data

This presentation outlines the plan for making a 2D flux map of the neutrino sky, via energy and direction reconstructions with transformer neural networks and flux estimations with unfolding.

Speaker: Lukas Liland

lamarr_lab_visits_pitch.pdf

30 Atmospheric Muon Flux from Stopping and High-Energy Muons in IceCube

Speaker: Lene van Rootselaar

31 Discussion & Buffer

16:00 Departure