Eligibility

  • Open to anyone attending Hackathon in person. RSVP via Luma: https://luma.com/xmn74jyd is required to enter the venue.
  • Solo participants and teams of any size are welcome. Judging is on the outcome, not the team size. If participated as a team, the prizes are for the team and NOT individuals.

What you build

  • You'll pick from a set of themes revealed in person. Picking more than one is fine, picking none is not an option. Your submission must clearly target a real problem within the chosen theme.
  • Elastic must be in the loop. Retrieval, context engineering, ingest, observability, ES|QL, vector search, Agent Builder, at least one Elastic component must do real work in your solution.
  • You must use real data. No toy datasets. Your agent must be grounded in sources you went and found.
  • Cite everything. Both in your write-up and in your agent's responses. Credit the original publishers.

What you can use

  • Any coding assistant, Claude, Cursor, Codex, Copilot, Kiro. Bring your own coding assistant.
  • Any Elastic feature on Cloud or self-hosted: Elasticsearch, Kibana, Elastic Agent Builder, Workflows, Open Crawler, Semantic Search, Knowledge Base, Jina embeddings via EIS, ELSER, ES|QL, Beats, Logstash, Elastic Agent.
  • Any GCP services you can stand up in the time available: Vertex AI (any model), Agent Engine, Compute Engine, Cloud Functions, Cloud Storage, Document AI, etc.
  • Open-source libraries, public APIs, and any data source whose terms allow your use case. Respect rate limits and ToS.

What you can't do

  • You can't ship something that was substantially built before the event. Reusing your own libraries, scaffolding, and personal templates is fine. Reusing a project you started last week is not.
  • You can't scrape sites in violation of their terms of service. If a portal forbids automated access, find another source.
  • You can't include personal data of real individuals without consent. Public data, scheme data, and aggregate statistics are fine; private individuals are not.

Originality and IP

  • Whatever you build, you own. Elastic and GCP/GDG Cloud New Delhi don't claim rights to your code or your demo.
  • You're responsible for any third-party code or data you include. License-incompatible reuse is your problem, not ours.

Judging

Area  
Real-world impact and relevance to New Delhi  
Data effort: scraping, cleaning, joining, enrichment  
Elastic usage  
GCP usage  
Actionability: alerts, summaries, escalation, recommendations the user can act on  
Demo quality and storytelling  

Decisions of the judging panel are final. Ties broken by demo quality.

Code of Conduct
Refer Elastic's Open Source Community Code of Conduct