We are a research group based at Aithyra in Vienna, Austria, led by Dr. Alex Tong.
Our group develops machine learning methods to model and engineer biological systems. We build foundational AI tools at the intersection of generative modeling, optimal transport, and geometric deep learning to understand the fundamental dynamics of biology with applications in single-cell biology and molecule design.
Explore our research, publications, and open positions on alextong.net and our Aithyra group page.
Our core mission is moving towards in silico biology, creating computational engines for molecular design through controllable generative modeling. Our primary research thrusts include:
- Generative Modeling & Flow Matching: Developing highly efficient, continuous-time generative models and simulation-free sampling techniques (e.g., Minibatch Optimal Transport, Flow Matching, Diffusion Models) that provide more flexible and efficient models.
- Generative Molecular Design: Creating more controllable models to design functional proteins and biologics from scratch. (Note: In a past life, Alex also co-founded Dreamfold to translate these challenges into real-world therapeutics).
- Decoding Cellular Dynamics: Applying flow models to understand how cells develop, respond to changing conditions, and how we can intervene to direct cell fate.
- AI for Science: Expanding our models to tackle broader scientific challenges, including crystal structure prediction and efficient Boltzmann generators.
Below are key repositories associated with our recent algorithms and papers:
| Project / Topic | Description | Paper/Venue |
|---|---|---|
transferable-samplers  |
Scalable, transferable Boltzmann Generators on Peptides | ICML 2025 NeurIPS 2025 |
Open-dLLM  |
Fully open source stack for training of diffusion langauge models | Incorporates PAPL ICLR 2026 (Oral) and P2 |
FoldFlow  |
Some of the original flow matching for protein design code | ICLR 2024 (spotlight), NeurIPS 2024 |
torchcfm  |
Implementations for our foundational work improving and generalizing flow-based generative models. | TMLR |
(Note: We actively maintain these repositories. Feel free to explore the code, open issues, or submit PRs!)
We are currently building out our team at the newly established Aithyra institute! We are looking for researchers who want to bridge fundamental AI theory with high-impact biological applications.
- Prospective PhDs & Postdocs: We have open positions in Generative AI, Multimodal ML, and AI for Life Sciences. Please reach out via email! PhD applications for Fall 2026 are closed at this time.
- Visiting Researchers: We welcome collaborations. Check out the contact instructions on Alex's website.
- Open Source Contributors: We welcome pull requests and issues on all of our public repositories.
📫 Contact: atong@aithyra.at
