Lessons learned from the 2025 agentic AI for science hackathon

Lee J, Neralla H, Rhys Campbell C, Zhang K, Ajith A, Ely J, Jiang H, Zhou Z, Ramlukan N, Li X, Thart Chiang H, Mahesh Kulkarni A, Thirumalaiswamy A, Pahi S, Habibur Rahman M, Pratoori R, Huang E, Iqbal Bhatti A, Prabhakaran P, Kumar M, Romero AH, Sun Y, Stiles C, Maheshwari C, Liu P, Shi L, Choudhary K (2026)

Publication Type: Journal article

Publication year: 2026

Journal

Machine Learning: Science and Technology Institute of Physics Publishing (IOP)

Book Volume: 7

Article Number: 040501

Journal Issue: 4

DOI: 10.1088/2632-2153/ae7f6a

Abstract

The rapid emergence of agentic AI presents new opportunities and challenges for accelerating scientific discovery through tool-augmented reasoning, autonomous workflows, and reproducible results. To explore these capabilities in a hands-on, community-driven setting, we hosted the Agentic AI for Science Hackathon 2025, which attracted 352 registered participants. The event was designed to benchmark, test, and extend agentic AI workflows across scientific tasks while remaining model-agnostic and lowering the barrier to entry for participants from diverse backgrounds. Participants engaged with a unified API ecosystem centered on the fully open-access AtomGPT.org API, implementing tool calling, asynchronous agents, and multi-model reasoning to solve problems spanning materials database retrieval, literature search, mathematical reasoning, and scientific workflow automation. The hackathon prompt emphasized reproducibility, transparent tool usage, and explicit agent-tool interaction, encouraging participants to move beyond single-prompt question answering toward structured, auditable pipelines. Outcomes included working agentic prototypes, identification of failure modes in contemporary chatbots, and actionable insights into best practices for agent design in scientific contexts. This paper documents the hackathon design, task structure, and key lessons learned, highlighting how hackathon-based evaluations can complement formal benchmarks and inform the development of robust agentic AI systems for science. Notebooks from participants who consented to public sharing are provided in the supplementary information. The GitHub repository for the hackathon is available at https://github.com/atomgptlab/aai4science.

Authors with CRIS profile

Sampanna Pahi Physics Underlying Life Science (PULS)

Involved external institutions

Johns Hopkins University (JHU) US United States (USA) (US) Arizona State University (ASU) US United States (USA) (US) Iowa State University of Science and Technology (ISU) US United States (USA) (US) Schrödinger GmbH DE Germany (DE) Vrije Universiteit Brussel (VUB) BE Belgium (BE) Hannam University / 한남대학교 KR Korea, Republic of (KR) Washington University in St. Louis US United States (USA) (US) West Virginia University (WVU) US United States (USA) (US)

How to cite

APA:

Lee, J., Neralla, H., Rhys Campbell, C., Zhang, K., Ajith, A., Ely, J.,... Choudhary, K. (2026). Lessons learned from the 2025 agentic AI for science hackathon. Machine Learning: Science and Technology, 7(4). https://doi.org/10.1088/2632-2153/ae7f6a

MLA:

Lee, Jaehyung, et al. "Lessons learned from the 2025 agentic AI for science hackathon." Machine Learning: Science and Technology 7.4 (2026).

BibTeX: Download