Overview
Build with DataHub: The Agent Hackathon invites participants to create AI agents that operate with deep data context. The event centers on using DataHub's open source Context Platform to empower agents with complete understanding of data assets, schemas, lineage, ML metadata, and governance. The goal is to demonstrate how agents can access and act on data in a way that reduces errors and saves time, ultimately turning a complex data stack into a practical and workable environment for autonomous or semi autonomous systems. The challenge is hosted publicly and online, welcoming diverse teams from around the world who want to prototype context aware agents that can read, interpret, and write back to a data governance graph.
What to Build
Submissions should leverage DataHub components such as MCP Server, Agent Context Kit, DataHub Skills, or Analytics Agent to create agents that understand data relationships and apply them to real tasks. Examples include agents that perform data quality checks, generate production code from schemas, or manage ML model lifecycles by tracing data from training to deployment. The emphasis is on context driven automation rather than simple data read operations. Submissions should clearly describe the problem solved, how DataHub is utilized, and the end to end workflow from input to output. A public project URL and a public code repository with clear setup instructions are required. A demo video under 3 minutes is expected to illustrate the project working in action. Optional sample outputs or artifacts may be included to showcase the solution.
Agenda & Format
The hackathon runs as an online event with a deadline to submit by August 10, 2026 at 5:00 pm Eastern Time. Participants are encouraged to provide a live demo or repository with setup steps, along with a narrative description of features and impact. The judging criteria cover how effectively DataHub is used, technical execution, originality, real world usefulness, submission quality, and potential open source contributions. Bonus prizes are available for meaningful open source contributions to DataHub projects or related documentation and for completing the Most Valuable Feedback Survey. The event also highlights resources such as blogs, guides, webinars, and help desks to support participants in building with DataHub.
Who should attend & Why
The challenge is suitable for data engineers, ML engineers, data scientists, product teams, and open source contributors who want to explore contextual agent development. It is designed for those who want to showcase how context aware agents can operate on data catalogs, lineage information, and governance rules. Participants gain exposure to DataHub capabilities, collaborate with peers, and have a chance to win cash prizes and recognition in the Devpost community. The event is online, allowing global participation, and emphasizes clear documentation, demonstrations, and reproducible results.
