my photo

Martin van der Schelling

PhD Candidate at TU Delft (Materials Science)

Hi! I'm interested in developing software for a better world, combining computational modeling and optimization with materials science. With my chemical background, I'm curious about the world of plastics and composites, and I'm ready to take on the challenges that come with the clean energy transition. I am a creative thinker and have a talent for making complex topics understandable. Besides, I play piano in a cover band!

Contact me :)

Experience

Delft University of Technology

PhD Candidate • Feb, 2022 — Present

PhD candidate in the Bessa Group on machine-learning-guided optimization for materials design.

  • f3dasm: Lead developer of f3dasm, an open-source Python framework for data-driven design of materials and structures with 120+ GitHub stars, 3,000+ monthly PyPI downloads and a JOSS publication, by designing a modular API that links design of experiments, simulation, machine learning and optimization, with MPI-parallel data generation on HPC clusters.
  • Python Code of Conduct: Co-author and maintainer of the Bessa Group Python Development Code of Conduct, the standard all of the group's Python developers follow to keep research code consistent, maintainable and FAIR, by writing its coding-style and packaging guidelines and an AI-agent review workflow that checks repositories against it automatically.
  • Learning to Choose Optimizers: Developing L2CO, a meta-learned optimizer-selection framework that matches or outperforms the single best optimizer on standard black-box benchmarks and was published at IEEE CEC 2026, by training a model that picks an optimizer from problem features alone and reassesses that choice during the run. Now extending L2CO to reinforcement-learning policies that switch optimizers multiple times within a single run.

Recent advances in computational resources have accelerated the development of inverse design approaches for structures and materials. In particular, data-driven strategies that leverage machine learning are increasingly shaping modern design workflows. However, constructing and maintaining large material-response databases remains challenging in practice. Typical challenges include data management, efficient parallel computing, and the integration of third-party simulation or analysis software. Because many applied fields remain conservative in openly sharing data and software, researchers often spend substantial time re-implementing standard procedures, hindering reproducibility, benchmarking, and adherence to FAIR principles.

In this work, we introduce f3dasm (framework for data-driven design & analysis of structures and materials), a general and user-friendly open-source package that supports the full data-driven design pipeline. The framework aims to streamline research workflows, facilitate the replication of existing studies, and encourage the sharing of new experiments and results. f3dasm emphasizes flexibility, interoperability, and modularity. In particular, it enables researchers to integrate different software packages and computational tools across the key stages of a data-driven design loop: (1) design of experiments; (2) data generation; (3) machine learning and surrogate modeling; and (4) optimization.

Since its first public release, f3dasm has been expanded to support MPI-based parallelization in high-performance clusters, enabling efficient large-scale data generation and simulation workloads. Furthermore, f3dasm adopts array-centric data structures that avoid breaking the computational graph when interfacing with automatic differentiation libraries, facilitating gradient-based learning and optimization. By lowering technical barriers and promoting reproducibility, f3dasm contributes to democratizing data-driven design for researchers and practitioners in structural and material engineering.

The data-driven design process supported by f3dasm
The four stages of the data-driven design process covered by f3dasm: design of experiments, data generation, machine learning, and optimization.

van der Schelling et al., (2024). f3dasm: Framework for Data-Driven Design and Analysis of Structures and Materials. Journal of Open Source Software, 9(100), 6912, https://doi.org/10.21105/joss.06912

Research software written during a PhD often outlives the PhD itself, yet in practice it is frequently left without tests, documentation or a consistent structure, which makes it hard for the next researcher to pick up, reproduce or build on. The Bessa Group Python Development Code of Conduct addresses this by setting one standard that every Python developer in the group follows, so that code can be handed over, shared between projects and published as open-source science following the FAIR principles (Findable, Accessible, Interoperable and Reusable).

The Code of Conduct separates what is required from how to achieve it. The guidelines are mandatory and cover project creation, a coding style, how to structure, document and publish a Python package, and a release every six months that is peer-reviewed by the group during a Code Release Week. The accompanying tool guides (Ruff, pre-commit, MkDocs and documentation hosting) are suggestions of how to comply, and a checklist summarises every requirement together with how it is verified.

To make that review cheaper and more consistent, I added an automated review workflow: a self-contained prompt that has an AI coding agent (Claude Code) install, lint, test, document and build a repository inside a sandboxed Docker container, and report its compliance with the checklist without modifying any code.

Optimization plays a central role in data-driven design. However, choosing an appropriate optimization algorithm for a specific task is often challenging, as it depends on the structure of the optimization problem and practical limitations such as the computational budget. To tackle this challenge, the field has seen a shift from manually designing optimizers toward using machine learning methods that attempt to learn the optimization process itself. Despite significant progress, existing approaches frequently exhibit limited generalization to new distributions and typically require retraining to transfer their knowledge.

We introduce a new approach called "Learning to Choose Optimizers" (L2CO). In our approach, a meta-learner first selects an optimizer before any function evaluations—based solely on available problem characteristics—and then reassesses that choice using a unified model that jointly encodes problem context, optimizer identity, and the observed optimization trajectory. By restricting selection to well-established algorithms, L2CO avoids the convergence issues of learned update rules while retaining the adaptability of dynamic selection.

In this work, we train our model offline on a collection of benchmark functions, employing a mixture of gradient-based, population-based, and probabilistic model-based optimizers. We demonstrate the effectiveness of our method by comparing its performance to classical optimizers on standard benchmark problems and a neural net classification task. The empirical results indicate that L2CO is a promising tool with improved generalization capabilities.

Overview of the L2CO framework
Overview of the L2CO framework. A first decision selects an initial optimizer from the problem features alone; a second, trajectory-aware decision then fuses problem embeddings, optimizer embeddings and the observed trajectory to either confirm that choice or warm-start a switch to a more suitable optimizer.

van der Schelling, M., Toshniwal, D., & Bessa, M. A. Learning to Choose Optimizers. IEEE Congress on Evolutionary Computation (CEC), 2026.

Brown University

Visiting Scholar • Sep, 2023 — Feb, 2024

Taught probabilistic machine learning to 30 graduate students in Brown's Data-Driven Design & Analysis of Structures and Materials course (3dasm), giving two guest lectures and supervising final projects. Contributed homework assignments, project exercises and lecture notebooks to the open-source course material.

ALTEN Nederland

Consultant Technology • May, 2021 — Feb, 2022

Consultant Technology at ALTEN Nederland, placed at KROHNE New Technologies.

  • Masterclass: Earned the ISTQB Foundation certificate and completed ALTEN's six-week C++ software masterclass by building an autonomous Bomberman agent with object-oriented design, UML and unit tests.
  • Portable clamp-on flowmeter: Automated testing of KROHNE's companion Android app by writing Xamarin UITest suites in C# and JavaScript scripts that verify the flowmeter's sensor data, and reported the results to the development team.
  • Foodflux: Built automated power-profiling tests for KROHNE's hygienic Foodflux flow sensor by scripting the measurements in JavaScript, analyzing them in Python and maintaining the Arduino C++ control of a servo potentiometer.

NPSP B.V.

MSc thesis student • Mar, 2020 — Mar, 2021

Designed bio-based composite recipes with improved mechanical properties over NPSP's existing bulk-moulding compounds, by developing an open-source Bayesian-optimization tool (BMC-optimizer) that suggests new natural-fiber weight ratios from measured recipe performance.

NPSP B.V. is a research and development company within the field of bio-based composites. New circular materials are being produced by using materials from waste streams.

Besides the Bayesian-optimization tool, I delved deeper into meta-heuristics and meta-learning. Based on the loss-landscape of general optimization problems, I have developed a supervised classifier that chooses the most competent algorithm for each problem.

Delft University of Technology

Teaching Assistant Computational Materials Science • Feb, 2020 — Apr, 2020

Assessed the weekly coding assignments of 80 master's students in 'Computational Materials Science I'. Reviewed their MATLAB and Python code with written feedback and published worked solutions as Jupyter notebooks.

V.O.F. De Klittenband

Manager & keyboard player • Sep, 2018 — Present

Run a student cover band as a small business with 30+ shows per year, alongside playing keyboard.

  • Clients & crew: Secure and manage the band's bookings by negotiating fees and terms with clients and aligning the technical crew on sound, lighting and stage requirements.
  • Logistics: Organize every performance end-to-end by planning transport, equipment, crew and band availability for each show.
  • Finance & administration: Automated the band's administration by building custom Python software that generates the quotation, contract and invoice for each gig and calculates its price from musician fees, equipment rental and travel costs. Also synced bookings from the CRM into a shared band calendar.

Show at Openingsfeest Delft
Show at Eurekaweek Rotterdam
Show at lustrum Virgiel

OGD ict-diensten

IT-support • Nov, 2016 — July, 2018

Provided first-line IT support to employees at Priva and Ampelmann, and improved internal IT procedures by working with Active Directory, TOPdesk, Windows 10 and Office 365.

  • Ampelmann B.V. (May, 2018 — July, 2018)
  • Priva B.V. (Nov, 2016 — Apr, 2017)

Education

Delft University of Technology

MSc Materials Science & Engineering • Sept, 2018 — Mar, 2021

  • MSc thesis: Developed a data-driven strategy that switches between optimization algorithms mid-run based on problem features, outperforming each individual optimizer on the test set and running 2.5× faster than Bayesian optimization, by training classifiers on the convergence behavior of a generated set of benchmark problems; graded 9.5/10.
  • Specialization: Graduated with a GPA of 8.5/10 in a self-defined specialization in plastics, sustainability and computational materials science, by combining courses in polymer science, functional coatings, self-healing materials, clean-energy materials and computational materials science.

With a background in chemical engineering, I am interested in the world of polymers and soft matter. I took courses in 'Functional Coatings' and 'Self-healing Materials' to become an expert in incorporating chemistry in modern material science. In addition, the minor 'Sustainable Energy and Technology' and the MSc course 'Materials for Clean Energy Technology' have taught me a lot about the world-wide drive of creating renewable energy and products. With my knowledge in chemical material sciences, I would like to be part of the global sustainability trend. Lastly, the course 'Polymer Science' has taught me that unconventional materials such as polymers are complex to model. Because of that, the multiple courses on Computational Materials Science are an essential aspect to my specialization.

Imperial College London

External Research Project • Sept, 2019 — Dec, 2019

  • Research: Simulated hole mobility in the conjugated polymer C16-IDTBT, reproducing the trend of experimental transient SCLC measurements, by writing a Python pipeline that maps molecular-dynamics morphologies onto the Kinetic Monte Carlo code ToFeT; graded 9.0/10.
  • Publication: Co-authored the PNAS paper 'Perpendicular crossing chains enable high mobility in a noncrystalline conjugated polymer' (2024), by contributing the kinetic Monte Carlo charge-transport simulations of C16-IDTBT.

To enhance the performance of organic solar cells, it is essential to investigate the microscopic features that affect the charge mobility in semiconducting materials. The co-polymer indacenodithiophene-benzothiadiazole (C16-IDTBT) has shown promising results for high mobility hole transfer. This research attempts to integrate the microscopic dynamics of this co-polymer onto the macroscopic simulation program ToFeT. The Time-of-Flight and field-effect transistor mobility measurement methods and their implementation in the simulation code are described. Subsequently, a network of inter- and intramolecular interactions is built, and the morphology of C16-IDTBT is extracted from a molecular dynamics simulation in Gromacs. The resulting hole mobilities are compared with transient SCLC and FET experimental data from the literature. The time-of-flight mobilities show similar behavior to experimental transient SCLC mobilities.

Snapshot of C16-IDTBT
Molecular structure of C16-IDTBT.
From this three-month internship, I have learned to work in an interdisciplinary team. With limited prior knowledge of C, I managed to understand and use a complicated Kinetic Monte Carlo program and adapt it to my own research. Furthermore, I worked with Python, bash and a cluster computer network.
Flowchart of the research project
Flowchart of the external research project.
This research was part of the EU-funded project CAPaCITy.

This work contributed to the following peer-reviewed publication:
  • J. F. Coker, S. Moro, A. S. Gertsen, X. Shi, D. Pearce, M. P. van der Schelling, Y. Xu, W. Zhang, J. W. Andreasen, C. R. Snyder, L. J. Richter, M. J. Bird, I. McCulloch, G. Costantini, J. M. Frost and J. Nelson, 'Perpendicular crossing chains enable high mobility in a noncrystalline conjugated polymer', Proceedings of the National Academy of Sciences 121(37), e2403879121 (2024).

Delft University of Technology

BSc Molecular Science & Technology • Sept, 2013 — Mar, 2018

  • BSc thesis: Characterized the linker rotational dynamics of UiO-66 metal-organic frameworks for molecular data storage, showing that UiO-66-NO2 matches the linker–framework interaction of MIL-53-NO2, by synthesizing and characterizing the frameworks (PXRD, DRIFTS, TGA, N2 adsorption) and measuring them with broadband dielectric spectroscopy.
  • Minor in Sustainable Energy & Technology.

Due to the ever-increasing demand for efficient data storage, innovative ways of storing data on the molecular level are being investigated. To store data on the molecular level, rotating linkers of the UiO-66 metal-organic framework (MOF) are possible candidates. Before we can control this rotation, the rotational dynamics of the 1,4-dicarboxylic linker are being researched. With the characterization techniques PXRD, DRIFTS, TGA and N2-adsorption, it was concluded that the structure and surface area are the same as those of reported materials. With broadband dielectric spectroscopy, the rotational dynamics of the metal-organic frameworks were examined. In addition, we compared UiO-66-NO2 with another MOF with the same BDC linker, MIL-53-NO2, and it showed that the dielectric relaxation is around the same temperature. This suggests that the interaction of the linker and the framework is of equal magnitude. In conclusion, UiO-66-NO2 has a similar interframework interaction to MIL-53-NO2. The large difference between relaxation temperatures for UiO-66-NH2 and UiO-66-NO2 may be explained by the interactions between the cornerstone and the linker which can hinder the rotation.

Broadband dielectric spectroscopy of UiO-66-NH2
Low-frequency broadband dielectric spectroscopy of UiO-66-NH2.
During my bachelor's thesis, I have gained hands-on experience in a chemical lab and can therefore work independently, safely and efficiently.

Associations

Faculty of Mechanical Engineering (ME), TU Delft

PhD council member • 2022 — 2024

Represented the PhD candidates of TU Delft's Faculty of Mechanical Engineering in the faculty PhD council, by raising PhD concerns with the faculty and organizing the council's sports event for the PhD community.

Hockeyclub Delfshaven

Competition Secretary • 2021 — 2026

Ran the weekly match program of field hockey club HC Delfshaven, by scheduling all home matches and appointing a referee to each.

Student Association 'Tubalkain'

Board of Studies member • 2018 — 2019

Represented the students of the MSc Materials Science & Engineering in safeguarding its educational quality, by collecting course feedback from students and raising it with the program's teaching staff.

K.S.V. Sanctus Virgilius

Advisory Board Theatre Production • 2018 — 2021

Guided three successive production boards on long-term policy, by advising them from my own experience of staging the 2016 musical.

K.S.V. Sanctus Virgilius

Board member Theatre Production • 2015 — 2016

Staged a student musical for 1,300 ticket-holders over three performances at Theater de Veste, Delft, with a cast and crew of 50 students and professionals, by managing the orchestra and the stage design.

AEGEE-Delft

Secretary Art & Culture committee • 2014 — 2015

Organized cultural activities for AEGEE-Delft, including the art and music festival 't Collectief, by acting as committee secretary and coordinating the committee's planning and communication.

Skills & Interests

Programming languages

Python, C++, MATLAB

Applications and tools

Git, LaTeX, Anaconda, Visual Studio Code, QtCreator, VideoPad Video Editor

Methods

Scrum, Object-Oriented Programming, Data-Driven Design

Licenses

PADI Open Water Diver, ISTQB Foundation Level

Field hockey

Field hockey enthusiast. I enjoy playing in Delft and Rotterdam, and played in London during my internship.

Playing piano

I enjoy making music with friends and performing with my coverband. Check out this livestream I took part in!

Social Links

  • Github: https://github.com/mpvanderschelling
  • LinkedIn: https://www.linkedin.com/in/mpvanderschelling/

Martin van der Schelling — martin@vanderschelling.com