Introducing The ASF’s New Logo Read Now

Board minutes, 2026-10-28

ASF Oak Leaf Icon

[raw] the published text, which is the record. [2026] [main index]

                    The Apache Software Foundation

              Special Board of Directors Meeting Minutes

                           October 28, 2026


 1. Call to order

    This is a special meeting of the Board of Directors, called under
    section 5.11 of the Bylaws of the Foundation.

    The meeting was scheduled for 22:00 UTC and began at 22:02 when
    a sufficient attendance to constitute a quorum was recognized by
    the chair.

            Other Time Zones: https://www.timeanddate.com/worldclock/fixedtime.html?iso=2026-10-28T22:00:00&msg=ASF+Board+Meeting

    The meeting was held via teleconference, hosted by the Secretary
    via Zoom.

    The #asfboard channel on the-asf.slack.com was used for backup.

 2. Roll Call

    Directors Present:

        Zili Chen
        Shane Curcuru
        Christofer Dutz
        Emmanuel Lécharny
        Justin Mclean
        Jean-Baptiste Onofré
        Greg Stein
        Sander Striker

    Directors Absent:

        Christopher Schultz

    Executive Officers Present:

        Craig L Russell
        Ruth Suehle

    Executive Officers Absent:

        none

    Guests:

        Paul King
        Brian Proffitt
        Andrew Musselman
        Sally Khudairi
        Jeff Genender
        Dave Fisher
        Jarek Potiuk
        Thomas Neidhart
        Henri Yandell
        Philipp Ottlinger
        Kanchana Welagedara

 3. Special Orders

    A. Increase Staffing Budget

       See Attachment A

       WHEREAS, the Board of Directors has determined that additional capacity is
       required to support the Foundation's operations; and

       WHEREAS, the Foundation is building an AI platform consisting of a governed
       gateway to language models, an agent workbench on which Apache projects can run
       agents without standing up their own infrastructure, and project-facing use
       cases including security scanning, repository health, and issue triage and
       patching; and

       WHEREAS, parts of that platform already carry real work on behalf of Apache
       projects, and the remaining work has been scoped and published as a work plan
       that exceeds the capacity available to the Foundation on a volunteer basis; and

       WHEREAS, the capacity contemplated to deliver that work plan is one platform
       lead and up to two platform engineers, engaged as independent contractors, as
       described in the summary accompanying this item; and

       WHEREAS, the Apache Software Foundation maintains a consolidated Staffing budget
       line in order to protect the privacy of individual compensation arrangements;
       and

       WHEREAS, the process of assembling and managing the Foundation's annual budget
       is directed by the President;

       NOW, THEREFORE, BE IT RESOLVED, that the Staffing budget line be and hereby is
       increased by $300,000 for the current fiscal year; and be it further

       RESOLVED, that this figure is set deliberately above the cost of the engagements
       contemplated, in order to leave headroom to extend, add, or re-shape an
       engagement as the work proceeds without returning to the Board, and that no part
       of this increase need be spent; and be it further

       RESOLVED, that the President is authorized and directed to incorporate this
       adjustment into the Foundation's budget in the ordinary course of budget
       assembly and management.

       Special Order 3A, Increase Staffing Budget, was approved (yes:
       Zili Chen, Emmanuel Lécharny, Jean-Baptiste Onofré, Greg Stein,
       Sander Striker; abstain: Shane Curcuru, Christofer Dutz, Justin
       Mclean).

 4. Discussion Items

 5. Adjournment

    Adjourned at 22:40 UTC


============
ATTACHMENTS:
============

-----------------------------------------
Attachment A: Increase Staffing Budget - Motivation

Refractory is the Foundation's AI platform. It exists because thousands of
Apache contributors are otherwise adopting AI tooling one personal account
at a time, with no Foundation-wide view of what is spent, on whose behalf,
or under what terms. It has four parts:

  * llm.apache.org (LLMAO), the LLM gateway. One governed path to every
      model the Foundation uses: a committer authenticates with their Apache
      identity, receives a credential scoped to a project, and the usage is
      metered and attributed to that project and that person.  Per-project
      budgets are enforced at the gateway rather than reconciled afterwards.

  * Gofannon, the agent workbench. Where agents are built and run,
      together with agent memory, so that work done once is available to the
      next run, and a workloads layer through which a project can run a job
      without meeting the agent machinery underneath.

  * Memory, durable storage for agent output and input, for recoverable,
      reusable, and traceable results.

  * Use cases delivered to projects in production or prototype: security
      scanning, repository health, issue triage and patching.

Already in hand: The workbench and the scanning pipelines are in use; the
gateway is a working proof of concept being taken to production.

Remaining, as published in the work plan: roughly 38 engineer-weeks of
committed work -- 11 on the gateway, 18 on the workbench and memory, 6 on
project-facing use cases, and 3 alongside Infrastructure -- with a further
13 weeks in the backlog. Every item names the architecture decision that
could change it, and both the plan and the decisions are public.

Headcount
---------

Role                   Count     Engagement
---------------------  -----     ----------------------------
AI Platform Lead         1       Contract, full-time
AI Platform Engineer   up to 2   Contract, full-time

AI Platform Lead. Technical lead for the platform, reporting to the Vice
President, Responsible AI. Owns the architecture, the public decision
record, the production operation alongside ASF Infrastructure, per-
project budgets and the spend reporting that follows from them, and the
relationships with peer foundations adopting the platform. Hires, manages
and mentors the platform engineers, and builds alongside them. Requires a
hands-on engineer who has shipped and operated production AI or data
platform systems, has led or managed engineers, and has an open source
track record of their own.

AI Platform Engineer. Builds and operates the two services and the
pipelines on top of them: budgeting, metering, routing and monitoring on
the gateway; self-serve agent pipelining with memory on the workbench; and
the project-facing use cases above. Requires documented experience as an
AI-focused software engineer, production CI/CD, Git, web development and
database experience, an open source track record, and comfort on a system
where the right answer is often to adopt the upstream rather than build
our own.

Both roles are independent contractor engagements, remote, with core hours
overlapping the team and no expectation of travel. US residents preferred.

Why $300,000
------------

Two engineer engagements at the intended band annualize to $300,000 to
$370,000. The lead engagement and the headroom account for the balance.
The engagements are expected to begin on a staggered basis, so the draw in
the current fiscal year is below the annualized run rate.

The remainder is deliberate rather than residual. The work plan is public
and will change as the platform meets real use; the headroom exists so
that extending an engagement, adding a third engineer, or funding a short
specialist engagement does not require a further resolution. Anything not
spent is simply not spent.


------------------------------------------------------
End of minutes for the October 28, 2026 board meeting.

Subscribe to ASF Plus One, Our Monthly Newsletter

Subscribe Now