Twitxy Saved

Yiming Su

@Yiming_Su3

Distributed Systems, SRE
PhD student @ UIUC @siebelschool, prev @MSFTResearch
Chambana yimingsu01.github.io Joined March 2026
180 Following 115 Followers 135 Posts

Latest posts 5

Yiming Su @Yiming_Su3
A little sneak peak of SREGym 2.0!
Jackson Clark Verified account @HacksonClark 2d
Did your email go down last week? This latest Microsoft outage was rather wild. It's starting to feel like the world is imploding. The central premise of #SREGym is to see if AI can stop such outages.

As a sneak peak of SREGym 2.0, we've recreated the Outlook outage. Check it out: github.com/SREGym/SREGym/… it's currently very challenging for frontier models.

What other outages would you like to see?

Stay tuned, we have something coming after Thanksgiving 🦃
View quoted post on X
Yiming Su @Yiming_Su3
Following up on the crazy idea in our first ever blog, we tested an SRE diagnosis pipeline **entirely driven by Jev**, with no LLM agent.

It passed 80/105 diagnoses (76.2%) across 21 fault scenarios with a median diagnosis time: **14.6 seconds**.

1/7
2 replies 2 reposts 10 likes 2.5K views View on X
Yiming Su @Yiming_Su3
SREGym is accepted in NeurIPS! There’s still MUCH work to be done on AI SRE, and a solid benchmark is always the first step.

It is also my great honor to be a co-lead of the project with @HacksonClark and work with the brightest team in the world. Let’s keep shipping!
Jackson Clark Verified account @HacksonClark Sep 25
I'm excited to share that SREGym.com has been accepted to NeurIPS 2026! 🎉

Huge thanks to my co-lead @Yiming_Su3, our co-authors @SaadMRP, @lilygn6, Yifan Tian, Hans-Arno Jacobsen, @yinfang_chen , @TianyinXu, and our 60+ contributors!

We have benchmarks for agents that write code.

We built one for what happens after you deploy it.

Can AI resolve production issues? 🧵 [1/N]
View quoted post on X
0 replies 0 reposts 11 likes 472 views View on X
Yiming Su @Yiming_Su3
Check out SREGym’s first ever blog post - and it’s on Jev!

Seeing Jev’s release, we immediately think that it can be extremely valuable to SRE agents. Check it out!
Jackson Clark Verified account @HacksonClark Sep 18
Can a small and fast decision model make SRE agents more reliable? 🤔
We gave an agent access to @typesafeai’s Jev through the Codex harness and tested it on 10 SREGym-Lite problems.
Three findings stood out. More details in the thread. 🧵
View quoted post on X
Yiming Su @Yiming_Su3
Check out SREGym (sregym.com)!
www.sregym.com SREGym An AI-Native Platform for Benchmarking SRE Agents
Snorkel AI @ NeurIPS 2026 Verified account @SnorkelAI Sep 8
Frontier Data Summit: Oct 8, SF. @fchollet, @sanmikoyejo, @nikogrupen, @ajratner, @alexgshaw, @StevenDillmann, and others onstage plus 25+ posters with the teams behind Agent's Last Exam, OSWorld 2.0, Terminal Bench Science, Terminal Bench, CollusionBench, PostTrainBench, T² Scaling Laws, SlopCodeBench, LONGRUN, SREGym, BizBench, and more.
View quoted post on X