• Machine learning for target discovery

    Applying representation learning and generative models to omics data to help prioritise and validate targets. This is the work I do with GSK.

  • Spatial and single-cell genomics

    Method selection, analysis design and interpretation for spatial transcriptomics and single-cell data, including niche identification, cell-cell communication and multi-sample integration.

  • Foundation models for biological data

    Pretraining, adapting and evaluating foundation models on omics data, and judging honestly where they earn their cost against a simpler baseline.

  • Research models into production

    Taking a model from a notebook to something a team can run: pipelines, MLOps, and platform work on Azure and Databricks.

Before and alongside the doctorate I spent seven years in data science consulting, most recently as an Advanced Analytics and Data Science Manager at Avanade. That included designing an Azure Databricks data science platform for one of the largest pharmaceutical companies, leading a team of ML engineers building production machine learning in Azure, and starting an MLOps practice across Avanade Germany, Austria and Switzerland.

The research side is a PhD in computational biology at Helmholtz Munich and the Technical University of Munich, first-authored work in Nature Genetics, and several open-source tools that other groups run. What I bring is the combination: methods that stand up to review, and systems that survive contact with real data.

If you have a problem in this territory, write to me at sebastian.birk@outlook.com. I am happy to say when something is outside what I can usefully help with.