PostDoc opportunity in AI for protein science

What we offer

• Join an established international collaboration between young PIs at the intersection of AI and experimental biology, with a proven track record of impactful research and extensive industry experience.
• Experience working at top research universities and institutes at the forefront of protein design — Technical University of Munich (DE), Duke (US), and Helmholtz (DE).
• Collaborate with experimentalists to learn about interesting biology and validate your in-silico work through wet lab experiments.
• Work on high-impact problems in protein design and synthetic biology, with insights from industry.
• Join a vibrant local ecosystem of labs, for instance through locally organised meet-ups like the Protein Design Stammtisch (DE) and the Triangle Protein Design (TriPoD) seminar series (US), with access to top-tier compute and lab resources.

What we are looking for

A curious, self-driven PhD with a background in computational biology, bio/physics or computer science, and:

• Strong experience in applied deep learning — for example, model training using PyTorch or TensorFlow.
• Understanding of current trends in AI: transformer models, GNNs, SE(3)-equivariant networks, diffusion models, and so on.
• Bonus: hands-on work with protein language models (e.g. fine-tuning), structural prediction tools (AlphaFold, ESMFold), or distributed training.

Interested?

Reach out with your CV and a short statement of interest, to msca at learning.bio. We will help develop a strong application for the MSCA Postdoctoral Fellowship.

In the 2025 call, we asked people to reach out by 15 July; the application deadline was 10 September 2025.

About us

Michael Heinzinger

Michael got his PhD (Dr. rer. nat.) in Bioinformatics (summa cum laude) from the Technical University of Munich in 2022. His thesis, How to Speak Protein? — Representation Learning for Protein Prediction, focused on adapting representation learning methods from natural language processing to protein sequences. Importantly, he was among the first to demonstrate the practical usefulness of the learnt protein representations for a variety of structural and functional protein features. The relevance of his thesis was honoured independently by being awarded among the finalists for the Deutsche Studienpreis, an award honouring the most influential dissertation within Germany every year. After finishing his PhD, Michael was among the first to expand the input repertoire of protein language models towards making protein 3D structures amenable to protein language models, rendering them multi-modal.

While working at Sanofi, Michael worked as a Computational Scientist within the newly formed Biologics x AI Moonshot (BioAIM) team, applying and developing predictive and generative AI approaches for biologics research, which led to a successful patent application.

Since March 2025, Michael leads his own team within the Institute of Computational Biology at Helmholtz Munich, and works as a lecturer at the Technical University of Munich.

Christian Dallago

Chris earned his PhD in Informatics (summa cum laude) from the Technical University of Munich in 2023. During his doctoral studies he made advances in bio-sequence representation learning, helping to establish the field — particularly through early work on transformer models for proteins and nucleotides. He played a key role in launching rigorous evaluation standards for protein models in design and engineering, introducing benchmarking datasets and data-splitting analyses.

As of 2025 Chris holds a dual appointment: Visiting Assistant Professor at Duke University, and leader of an applied research group in Digital Biology at NVIDIA. At Duke he focuses on exploratory research in biological machine learning; at NVIDIA he leads applied research in accelerated drug discovery. More on the team page.