Evolved 2025: Hackathon for Accelerating AI Breakthroughs in Healthcare & Life Sciences
Researchers and startups looking to computational power, biology, and medicine are converging faster than ever, Evolved 2025 emerges as an exciting global hackathon series designed to ignite AI-driven innovation in biotech and clinical development. Currently in its second years, Evolved is organized by Nucleate and supported by Nebius, Lux Capital, and NVIDIA, this program seeks to bring together software engineers, AI researchers, and biotech innovators to tackle some of the most pressing challenges in healthcare and life sciences by way of cutting-edge computational tools.
What is Evolved 2025?
Evolved 2025 is structured to maximize both local and global impact, combining regional and virtual hackathons with a global finale, with both regional and global hackathons allowing free 24 hours of access to H100 GPUs and pre-trained models via the Nebius cloud computing platform Key elements include:
- Regional hackathons: (from October 2025) held in cities such as San Francisco and Boston: local scientists, engineers, and innovators join forces to address biotech and healthcare challenges specific to their region.
- Global virtual hackathon: (from Nov 7th to Nov 16th) open to teams anywhere in the world to participate remotely and work on their ideas for an AI-driven project development in therapeutic design, diagnostics, drug discovery, and more.
- Grand challenge showcase: hosted by NVIDIA, where the top innovations coming out of Evolved 2025 will be highlighted and promoted.
Participants will be challenged to bring computational creativity to domains such as protein design, molecular modelling, predicting clinical outcomes, synthetic biology, and other intersections of AI & bio.
Why it matters
Evolved 2025 comes at a moment when AIxbio is no longer fringe: breakthroughs in protein structure prediction, generative design, and clinical data modelling are moving fast and AGI is making inroads in not just science communication but even in theorising new cures and discoveries, grant applications and reviewing paper submissions. Yet, many challenges remain — safety, robustness, generalization, translational relevance, and bringing computational tools into real world therapeutic pipelines.
By combining hackathon energy, global reach, and biotech domain constraints, Evolved 2025 has the potential to seed projects that don’t just win prizes, but move toward real deployment.
Evolved 2025 will challenge participants across three main tracks:
- Agentic Automation and Workflow Optimization – building AI systems that collaborate with scientists and clinicians to streamline research and operations.
- Modeling Biological and Biomedical Systems – advancing methods for representing and inferring complex biological or clinical processes, often through multimodal or foundation models.
- Molecular Optimization and Genomic Modeling – applying computational tools to genomic and molecular data for drug development, biomarker discovery, and therapeutic design.
- Binder Design Challenge: One of the most exciting additions to this year’s program is the Binder Design Challenge. Participants will be tasked with developing a binder that is cross-reactive for both a human receptor and its mouse counterpart in the same receptor family. Submissions will be evaluated and screened by Adaptyv Bio, with the top three designs awarded Adaptyv Bio lab credits — giving winners the opportunity to take their computational breakthroughs into the wet lab for validation.
What we learnt in Evolved 2024: Code, Compile, Cure
To understand what’s possible, it helps to look at what emerged from last year’s edition: Evolved 2024
Some highlights:
| Team | What they did | What their work could lead to / Use-cases |
|---|---|---|
| EvoCapsid (1st Place) | Using the ESM3 (a large protein model) to redesign gene delivery proteins & RNA editing proteins with lower immunogenicity. | Safer gene therapy, reduced immune responses, better delivery mechanisms. |
| MSEffect (2nd Place) | Combined probabilistic models and MS/MS transformer models to predict spectra of unknown compounds. | Illuminating “dark chemical space,” discovering new small molecules, drug leads, chemical diagnostics. |
| GenPlasmid (3rd Place) | Fine-tuned generative models to design more effective promoters and genomic parts. | Designing synthetic biology circuits, better control of gene expression, more predictable biological engineering. |
| Brainstorm Therapeutics (Special Mention) | Developed a midbrain organoid platform from Parkinson’s patients and used NVIDIA BioNeMo / foundation-model-based network analysis protocols to model disease and test perturbations. | More realistic disease modeling, high throughput screening for neurodegenerative disease, and possibly drug discovery. |
What are the 2024 winners doing now?
While the full trajectories are still emerging, some of the teams from Evolved 2024 have begun pushing beyond prototyping toward more mature applications:
- They are exploring partnerships and proof-of-concepts (for example, using laboratory validation or collaboration with biotech firms).
- Some are refining their models (improving safety, reducing error, better validation).
- Others are looking at commercialization routes, licensing, or spinning out as startups.
Broadly, the momentum from 2024 shows that hackathon projects in AI & bio can lead to serious follow-ups, not just theoretical work.
What to expect from Evolved 2025
Based on Evolved 2024 plus present trends, here are some things participants and observers should look out for:
- Improved rigor & translational focus: Projects that not just develop novel computational methods but also think about real biological validation, safety, regulatory considerations.
- More cross-disciplinary teams: Engineers, computational biologists, clinicians, domain experts working together.
- Diversity of challenge spaces: Beyond protein design and small molecules, expanded into diagnostics, large molecule therapeutics, data science for clinical outcomes, synthetic biology, and possibly deployment in resource-limited settings.
- Infrastructure support: Mentorship, access to computational resources (GPUs/TPUs), data sets, even wet lab validation if possible.
- Post-hackathon continuity: Better pathways for winners and promising teams to continue work — seed funding, incubation, collaborations with pharma or biotech, etc.
How to get involved or support
- If you’re a researcher, engineer, student, you could form or join a team for a regional (United States Only) or the global virtual hackathon.
- If you’re from industry, you can mentor, provide data or resources, or sponsor challenge tracks.
- Others might support via funding, infrastructure, or even helping translate hackathon outputs into real clinical or therapeutic trials.
If interested do check out and apply here
