Advertisement · 728 × 90

Posts by kilian

Preview
HPC Systems Administrator (Hardware & Infrastructure Operations) in Business Affairs: University IT (UIT), Stanford, California, United States Please note: Visa Sponsorship is not provided for this position. The Sherlock HPC cluster is the flagship of Stanford’s research computing...

#Stanford Research Computing is hiring an HPC Systems Administrator to steward our physical infrastructure. 2,000+ compute nodes, GPU galore, IB islands, PBs of flash storage. None of it runs without you.
careersearch.stanford.edu/jobs/hpc-sys...
#HPC #datacenter #hiring

3 hours ago 1 1 0 0
Preview
The DOJ is targeting Stanford Medicine's nonexistent problem Akeke and Zhang defend against the DOJ's investigation into Stanford Medicine and advocate for the importance of diversity consideration in admissions.

Stanford medical students speak up on the DoJ investigation into admissions:

“We want to see Stanford enroll more Black, Latino and Indigenous medical and PA students, not for optics, but because they deserve to be here without having their success dismissed as the product of DEI.”

2 days ago 38 18 0 0
Preview
Storage Architect / Senior Storage Systems Administrator (Research & HPC Data Platforms) in Business Affairs: University IT (UIT), Stanford, California, United States Stanford’s Research Computing team is seeking a storage expert to join our Research & HPC Data Platforms group. This is a flexible-level posting;....

#Stanford Research Computing is hiring a Storage Architect / Senior Storage Sysadmin.
100PB+ of research data, 5+ billion files, Lustre/MinIO/HSM. The research is Nobel-caliber, the storage should be too.
careersearch.stanford.edu/jobs/storage...
#HPC #Lustre #Linux #hiring

3 days ago 1 1 0 0
Preview
Stanford: Protect student privacy — reject DOJ demands for admissions data The Trump Administration is attempting to coerce Stanford Medical School into releasing sensitive admissions data that could compromise student privacy. While Stanford is not yet under a court order t...

Urge Stanford to reject the DOJ's demands for sensitive personal information about Stanford Medical School students:

Please sign & share the petition:
alumni.standforcampusfreedom.org/petitions/ca...

6 days ago 50 23 1 2
Preview
Principal Storage Architect & Team Lead (Research & HPC Data Platforms) in Business Affairs: University IT (UIT), Stanford, California, United States We are seeking a world-class technical leader to oversee our primary research storage platforms. These services span the gamut from the 15PB...

#Stanford Research Computing is hiring a Principal Storage Architect, to lead our HPC Data Platforms team.
100PB+ of research data, full-flash file systems, Lustre at scale.
Come architect storage for Nobel-caliber research!
careersearch.stanford.edu/jobs/princip...
#HPC #Lustre #storage #hiring

1 week ago 0 2 0 0
The machines are fine. I'm worried about us. On AI agents, grunt work, and the part of science that isn't replaceable.

ergosphere.blog/posts/the-ma...

2 weeks ago 0 0 0 0

And it's not a ban, it's an 18-month pause, until Nov. 2027, to "assess the impact of such development on the environment and electricity grid".

2 weeks ago 0 0 0 0

For sure! But in a world where login nodes already have GPUs, that's even cheaper than buying a subscription.

3 weeks ago 1 0 0 0

Unless OpenClaw (or any agentic tool) helps them improve their HPC workloads and run them more efficiently? There could be a benefit in spending some GPU dollars now to allow more jobs to run on the same amount of GPUs in the end.
And we're definitely seeing an trend to experiment in that direction.

3 weeks ago 0 0 1 0
Advertisement

Login nodes with GPUs? Ouch! That's wild.

3 weeks ago 0 0 0 0
screenshot of a resignation letter expressing sadness at leaving NASA

screenshot of a resignation letter expressing sadness at leaving NASA

My resignation letter

4 weeks ago 3933 1016 80 85

The nature of HPC environments with their job schedulers, resource usage policies and time limits are not the best setup for persistent LLM endpoints.
Creative workarounds are always possible, but there's probably even less incentive to hijack GPUs for OpenClaw than during the crypto mining boom.

4 weeks ago 0 0 0 0

Not OpenClaw specifically, but we’re definitely seeing increasing use of Ollama for running local, open models privately. In addition to the rapid growth in adoption of commercial models, with Claude standing out as a clear leader.
And that's not just students. :)

4 weeks ago 0 0 1 0

“i’m still punk” i whisper to myself as i turn down the music in my car so i can see better

1 month ago 3704 766 7 0
Preview
Scratch that: Sherlock now has 15 PB of flash - Sherlock changelog We are excited to announce a major expansion of Sherlock's /scratch file system. We are adding 5 PB of full-flash storage, bringing the total capacity from 10 PB to 15 PB. This investment directly add...

Just expanded Sherlock’s /scratch to 15 PB of all-flash Lustre at Stanford Research Computing. Built in-house on community Lustre with our own backported patches, OSTs added live. No vendor black box. No appliance magic. Just open source and a couple people who know their way around it! 🧱🔧⚡️

1 month ago 4 1 0 0
I spoke to AI agent Claude
I spoke to AI agent Claude YouTube video by Senator Bernie Sanders

I have a feeling will be watching this a couple of times in the weeks, months, and years to come.

www.youtube.com/watch?v=h3At...

1 month ago 1 1 0 0
Preview
Nvidia removes Rubin CPX accelerators from its roadmap — Groq 3 LPUs take center stage as CPX is removed Groq will take the role?

And more about it: www.tomshardware.com/pc-component...

1 month ago 1 0 0 0
Preview
NVIDIA Vera Rubin Achieves 40 Million Times More Compute In 10 Years: 288 GB HBM4, 22 TB/s Bandwidth, 50 PFLOPs of AI Horsepower NVIDIA has officially unveiled its next-gen AI data center platform called Vera Rubin, powered by the Rubin GPU and Vera CPU architectures.

Yes, looks like it's been Groq'ed and branded as LPxx

> The Groq 3 LPX compute tray is comprised of 8 GROK LPUs and is codenamed LP30. Each Groq 3 LPU offers 500 MB of SRAM, 150 TB/s of SRAM bandwidth, and 1.2 PFLOPs of FP8 performance. Each chip houses 98B transistors.
wccftech.com/nvidia-vera-...

1 month ago 0 1 0 0
Advertisement
Preview
Free and private AI-assisted coding on Sherlock - Sherlock changelog We're excited to announce a brand new section section in the Sherlock documentation, covering how to bring AI coding tools into your HPC workflows, for free, and without sending your code and data any...

Tired of watching your API credits disappear mid-session? New Sherlock docs cover running Zed + Ollama entirely on-cluster: full AI coding assistant, free GPU allocation, zero data leaving Stanford.
news.sherlock.stanford.edu/publications...
#Sherlock #HPC #Stanford

1 month ago 2 1 0 0
Preview
Sam Altman says OpenAI shares Anthropic's red lines in Pentagon fight But he still wants a Pentagon deal.

Looks like OpenAI is siding with Anthropic on this: www.axios.com/2026/02/27/a...

1 month ago 1 0 0 0
Preview
FAQs for This Weekend’s Bad Bunny Concert Featuring Football What time is kickoff? The Bad Bunny concert kicks off around 8:00 or 8:30 p.m., but there will be pre-concert entertainment starting at 6:30 p.m. f...

Q: Why will football be featured during the Bad Bunny concert?*

A: The Bad Bunny concert is fantastic exposure for the NFL and a once-in-a-lifetime opportunity to put the game of American football in front of fans of the biggest musical artist in the world.

2 months ago 430 133 5 15

Yet www.merriam-webster.com/dictionary/f...
Dunno, fitting, isn't it? 😁

2 months ago 0 0 0 0

Photoshopped selfies would still work better than AI generated avatars, IMO :)

3 months ago 0 0 1 0

What's even the point? VAST doesn't have a creative team capable of taking employees' pictures?

3 months ago 0 0 1 0
Post image Post image

Cow Tools!

We have lived alongside cows for nearly 10,000 years.
We breed them and exploit them

It is now, only now, that we have discovered THEY CAN USE TOOLS

Here I describe our study

(paper) www.sciencedirect.com/science/arti... in @currentbiology.bsky.social
with @auersperga.bsky.social

3 months ago 1321 539 26 109
Advertisement
Post image

notice to Americans: the French are rapidly innovating in the burger space

3 months ago 38 3 5 0
Preview
Nvidia's basic context memory extension infrastructure In an effort to get to know the basic Nvidia KV cache extension infrastructure, the ICMSP (Inference Context Memory Storage Platform) better, we asked Nvidia some questions about the Vera Rubin Pod ra...

blocksandfiles.com/2026/01/12/n...

3 months ago 0 0 0 0

Unbelievable. This would be a terrible blow to American science, writ large. It would decimate not only climate research, but also the kind of weather, wildfire, and disaster research that has underpinned half a century of progress in prediction, early warning, and increased resilience.

4 months ago 4292 1908 103 83

If you are (as I am) watching The Princess Bride for REASONS today, please take a look at the quoted thread for a master's look at the masters of swords

4 months ago 28 21 0 0
Preview
Immigrants Once Avoided Some Regions of America. That’s a Big Reason We’re So Divided. Today’s debates over immigration and assimilation have their roots in patterns of settlement that go back centuries.

www.politico.com/news/magazin...

4 months ago 1 0 0 0