Build with DataHub: The Agent Hackathon

Overview

The Build with DataHub: The Agent Hackathon challenges participants to create AI agents that leverage DataHub's open-source context platform to solve real-world data problems. The hackathon emphasizes the importance of providing AI agents with complete context, including schemas, lineage, ownership, and ML metadata, to prevent hallucination and improve efficiency in data-related tasks. Participants are invited to build on DataHub's foundation, which is used by companies like Apple, Pinterest, and Netflix. The goal is to demonstrate what is possible when AI agents are equipped with comprehensive data understanding. This can include developing autonomous agents, generating production data code, protecting ML models, or exploring entirely new applications. DataHub provides essential tools for agents, such as the MCP Server, end-to-end ML lineage, and DataHub Skills, which enable direct access to catalog workflows. These components allow agents to interpret and interact with the modern data ecosystem effectively. The hackathon encourages innovative solutions that empower agents to perform meaningful work within data stacks. Resources like documentation, SDKs, sample datasets, and starter kits are provided to assist participants in building agents that can ship and make a tangible impact. The hackathon offers a platform to showcase the potential of context-aware AI in data management and operations.

What to build

Create a working software application that uses DataHub to solve one of the following challenges: - Agents That Do Real Work: Build AI agents that read DataHub through the MCP Server or Agent Context Kit to understand connections, take action, and write results back to the graph. - Metadata-Aware Code Generation & Development: Build agents that generate production data code (transformation models, pipeline DAGs, ingestion scripts, configurations, migration code) using DataHub Skills or the MCP Server for accurate schemas, lineage, and rules. Submissions should include sample generated artifacts. - Production ML Agents: Build agents for ML teams that protect models in production by using DataHub's end-to-end ML lineage (accessed via the Agent Context Kit or MCP Server) to catch problems before they impact ML systems. - Open / Wildcard: Build anything creative that uses DataHub as its foundation, utilizing any part of DataHub's open-source stack (MCP Server, Agent Context Kit, DataHub Skills, Analytics Agent, or other DataHub products).

Prizes

Key dates

• Registration & Submission Period BeginsJuly 06, 2026 at 9:00AM EDT
• Feedback Period BeginsJuly 06, 2026 at 9:00AM EDT
• Free Build Session: From Zero to a Working DataHub Agent in 30 MinutesJuly 21, 2026 at 10:00AM ET / 2:00PM UTC / 7:30PM IST
• Registration & Submission Period EndsAugust 10, 2026 at 5:00PM EDT
• Feedback Period EndsAugust 10, 2026 at 5:00PM EDT
• Judging Period BeginsAugust 17, 2026 at 10:00AM EDT
• Judging Period EndsAugust 31, 2026 at 5:00PM EDT
• Winners AnnouncedSeptember 08, 2026 at 2:00PM EDT

Eligibility

The Hackathon is open to individuals who are at least 18 years old or have reached the age of majority in their country of residence. Teams of eligible individuals and organizations (including corporations, non-profits, LLCs, partnerships) existing at the time of entry are also eligible. An individual can join multiple teams or enter individually while part of a team. If entering as a team or organization, a Representative must be authorized to act on their behalf. The Hackathon is not open to residents of or organizations domiciled in countries, states, provinces, or territories where prohibited by US or local law (e.g., Brazil, Quebec, Russia, Crimea, Cuba, Iran, North Korea, and OFAC-designated countries). Promotion Entities, their employees, representatives, agents, immediate family/household members, Judges, and their employers or affiliates are ineligible. Any individual or organization whose participation would create a conflict of interest, as determined by the Sponsor or Administrator, is also excluded.

How to enter

1. Visit datahub.devpost.com and click "Join Hackathon".
2. Register for the Hackathon by signing up for a free Devpost account or logging in with an existing one.
3. Spin up DataHub locally using the Quickstart Guide.
4. Develop a working software application that uses DataHub to address one of the defined Challenge Categories, incorporating the open-source platform with the MCP Server, Agent Context Kit, DataHub Skills, or Analytics Agent.
5. To be considered for the Most Valuable Feedback bonus prizes, register for the hackathon and complete the online feedback form during the Feedback Period (July 6, 2026 – August 10, 2026).
6. Complete and enter all required fields on the "Enter a Submission" page of the Hackathon Website during the Submission Period.

Submission requirements

• A working software application (Project) that uses DataHub to solve one of the specified challenges, incorporating DataHub via the MCP Server, Agent Context Kit, DataHub Skills, or Analytics Agent.
• A URL to your Project providing easy access for judges to test its functionality (live demo, hosted app, or repo with clear setup instructions).
• A URL to your public code repository containing all necessary source code, assets, and full instructions for the project to be functional. The repository must be public, open-source, and include an Apache 2.0 open-source license file visible at the top of the page.
• A text description summarizing your Project, including its features, functionality, technologies, and data used.
• A demonstration video of your Project, under 3 minutes, uploaded to YouTube, Vimeo, or Youku with public visibility enabled. The video must show the Project functioning and in action, and not include unauthorized third-party trademarks or copyrighted music/material.
• Optional: Include Sample outputs (e.g., an examples/ folder in your repository) if your Project generates artifacts such as code files, queries, reports, or transformations.
• Projects must be newly created during the Submission Period. Disclosure of any pre-existing code, frameworks, libraries, starter templates, or AI coding assistants is required.
• If using third-party SDKs, APIs, or data, entrants must be authorized to use them according to their terms and conditions or licensing requirements.
• The Project must be capable of successful installation and consistent running on its intended platform, functioning as depicted in the video/description.
• All Submission materials must be in English or include an English translation.
• The submission must be the original work of the entrant, solely owned by them, and not violate any intellectual property rights or other rights of any other person or entity.

Judging criteria

• Use of DataHub: How meaningfully the project uses DataHub's context graph, MCP Server, Agent Context Kit, DataHub Skills, or Analytics Agent. Strong submissions should contribute back to the graph.
• Technical Execution: Quality of implementation, robustness, and whether the project functions end-to-end as claimed.
• Originality: Creativity and novelty of the approach, going beyond existing DataHub features. Building on, extending, or composing existing features is encouraged; rebuilding is not.
• Real-World Usefulness: Clear value for a real data, ML, or AI platform team, solving a practical problem practitioners face.
• Submission Quality: Quality of the demo video, written description, and README, ensuring judges can understand the project, its importance, and find clear setup instructions.
• Bonus: Meaningful Open-Source Contribution: Submissions including meaningful open-source contributions to DataHub (new connectors, skills, fixes, RFCs, documentation improvements) will be looked upon favorably. Existing contributions extended for the hackathon also count.

Sponsors & partners

Resources

• DataHub Docs
• Quickstart Guide
• DataHub Skills
• Agent Context Kit
• DataHub MCP Server
• Analytics Agent
• DataHub Core Repository
• DataHub Skills Repository
• NYC Taxi Sample Dataset
• Healthcare Sample Dataset
• Fiction Retail Sample Dataset
• Join DataHub Slack (#agent-hackathon channel)
• DataHub Town Halls
• Hackathon Official Rules
• Register for Free Build Session (July 21)