Approve the NVIDIA-supported Fuzzball hackathon at SC - with a Wolford engineer on site a day early and a month of hardware notice from NVIDIA

Situation

In the L/P Quick Sync on 2026-08-18, Lindsay Aamodt brought the NVIDIA hackathon she has been building - in person in Chicago, Friday November 13 kickoff, build Saturday and Sunday, judging Monday, winner presented in the booth Tuesday as SC opens - and asked Peter to name the risks that would make it a flop in person. Peter approved it and attached conditions: someone who works for Chris Wolford goes on site a day ahead to stand the clusters up, explicitly not Lindsay; and NVIDIA must confirm exactly what hardware they are providing a month in advance so the networking is handled beforehand. He committed himself as a judge alongside Bjorn, Greg and Scott, and Wolford as mentor staying through Saturday.

Reasoning

Peter judged that the technology is not the risk - he runs Fuzzball on two Sparks on his own desk and is not a Fuzzball developer - but that an unattended install is. He said plainly that it will fall apart and break and that he does not know how or where, which is why the mitigation is a body on site for a day rather than a better script. Assigning it to a Wolford engineer rather than Lindsay is about who can debug an unknown failure, and he said so unprompted so it would not read as a comment on her. The month-of-notice requirement converts an external dependency into a dated one, so a late NVIDIA hardware decision surfaces as their slip rather than a CIQ scramble. He committed his own and Wolfords time readily because the events payoff is product-market research on what else can be built on the Spark.

Additional Context

NVIDIA will supply and set up the clusters, provide a mentor and a judge, and promote the event on Luma, but will describe themselves as supporting rather than sponsoring or partnering. Ten teams of four clusters was the working shape. The five tracks came from ideas submitted by booth passers-by at ISC and PEARC competing to win a Spark. Lindsay plans targeted promotion into the national labs. The Nvidia DGX Spark deliverable sits at 95 percent confidence in the same weeks TPS report.

Observed Evidence

Lindsay: But is there anything like big? Because we were trying to think through all of the risks that would basically make this a flop in person. Peter named the install as the risk and staffed it. Peter: I am running Fuzzball on two Sparks here on my desk and I am not a Fuzzball developer and I set it up. It is really easy at this point. There is a couple gotchas, you know, but between now and November, no, this will just be a script that you run. Peter on the judge and mentor roles: It sounds wonderful ... Whenever you need me there. Wolford will be there whenever you need him there.

Matching Patterns

Confidence Breakdown

Reasoning Depth Analysis