Atom Grants
Discover
  • Dashboard
  • Search
  • Deep Research
  • Favorites
  • Projects

    TelcoRetrieve Challenge: Unlocking Telco Knowledge by ITU

    TelcoRetrieve Challenge: Unlocking Telco Knowledge by ITU supports AI competitors improving a standards-based retrieval pipeline to help telecoms professionals get accurate, token-efficient answers.

    OverviewSimilar Grants
    OverviewSimilar Grants

    Funder: Zindi

    Due Dates: November 13, 2026 (Launch) | January 6, 2027 (Enrolments close) | January 13, 2027 (Submissions close) | January 13, 2027 (Submission selection closes) | January 30, 2027 (Reveal)

    Funding Amounts: $20,000 USD prize pool; five prizes from $1,500 to $8,000

    Summary: An AI competition rewarding more accurate and token-efficient answers grounded in telecommunications standards using TelcoRetrieve.

    Key Information: First-, second-, and third-place winners must assign worldwide copyright in their winning solution code to Zindi.


    Description

    This ITU challenge, hosted on Zindi, asks participants to improve TelcoRetrieve, an open, vendor-neutral retrieval-augmented generation (RAG) pipeline developed by GSMA. The pipeline retrieves information directly from telecommunications standards and cites source documents, addressing inaccurate citations, confusion between releases, and outdated model knowledge.

    Participants can improve retrieval through chunking, enrichment, hybrid search, and reranking, or fine-tune the generator and embedding models within the challenge’s model constraints. Evaluation prioritises standards-grounded answer accuracy and token efficiency, with public testing followed by held-out private evaluation. The goal is to help engineers, vendors, and standards bodies navigate specifications, check implementations, support interoperability testing, and reduce manual cross-referencing across the telecoms ecosystem.


    Atom

    See the full grant listing

    Sign in to view full eligibility details, sources, similar grants, and AI-powered analysis.