David Justo

The official home of the Python Programming Language

TLDR; The PSF has made the decision to put our community and our shared diversity, equity, and inclusion values ahead of seeking $1.5M in new revenue. Please read and share. pyfound.blogspot.com/2025/10/NSF-...
🧵

vibe: instrumental heavy metal while debugging 80s cursed C (crying encouraged)

Days since Peruvian president impeachment: 0
As of today, number of presidents since 2016: 9
Tomorrow: 10?

My sibling team that owns the core passwordless systems in Windows is hiring an SDE II. Great team, interesting projects.

jobs.careers.microsoft.com/global/en/jo...

🔁reposts appreciated. Do you know any tgirls who have gone to a US consulate visa interview under the current admin?

I'm british n have an extension to stay in the US, but my H1B stamp is expiring in October, so if I want to be able to travel I will need to leave the US.

Worried won't get back in

my power only grows from here.
Hyped about being able to merge into llvm more easily. Now, onto improving the windows Asan docs.

Summer Undergraduate Internship - reposts welcome!

Are you a senior undergrad, interested in Programming Languages? Do you want to visit Canada for a paid 12-week internship?

This stuff is exhausting

ASan is like malware that's good for you(r C/C++), don't let anyone tell you otherwise...

Gave a small presentation today at the Seattle llvm meetup about some of the stuff I've been working on within msvc. Had fun and got to rock the win95 shirt, good vibes :)

Do they have an online store by any chance? This is super cool

Holy shit

Curious if there's any specific pain points you can share. Would be useful feedback.

these challenges seem like a great fit for durable execution frameworks a la durable functions, temporal, etc.

Thanks for clarifying. Understood, and agreed.

Just trying to understand - what does "meaningful language" mean here?

From there it's not too hard to see how these billion param models trained on the entire Internet would produce remarkable human like response, despite being, at its core, just an algorithm

I think it helps a lot to be familiarized with more rudimentary NLP models (even as simple as "bag of words" models) and see that they perform remarkably well in small datasets.

New blog post: I am a Rust compiler engineer looking for a new job

nnethercote.github.io/2025/07/18/l...

Have you seen their work in the automated reasoning group? It's not lean per se, but it's formal methods work so it's in the same ballpark. I'm a fan.

Thank you for all your work!!

Kate Efimova's Resume Public Kate Efimova (925) 326-**** kate.efimova0@gmail.com Bay Area, CA LinkedIn GitHub Twitter Experience (5 yrs) Software Engineer II @Hint Healt...

✨ HIRE ME ✨

I'm a Full Stack software engineer with 5+ years of experience, looking for my next opportunity. My personal values include psychological safety, D&I, and direct communication.

Please share for reach <3

docs.google.com/document/d/1...

Ah, LNK1377: the containing library is corrupt.

Ah yes, ah yes. We meet again.
(cries)

Seattle fact - Best tsukemen at menya musashi fyi. That spicy curry tsukemen hits like a (good) truck

I think at a deeper level, all tech discourse is kinda broken. There's too much money to be made, so the hype waves come way too strong, the claims are always about "this is just like the new internet/iphone" and I think folks are just fatigued.

I share the sentiment. Just to add: I think the AI discourse also needs to be understood in the context of the recent hype waves: crypto, metaverse, web3, etc. I think folks are understandably primed to be skeptical.

So the agent is usually in a unutilized vm. But obviously, this is a very expensive way to go about it. I just haven't had to optimize it because I already had these vms. (2/2)

Ah, this is where I reveal a pretty deranged part of my setup. I have 3-4 cloud VMs that I use for multi-tasking projects. They all host the same repo, but at different states. This just allows me to run heavy work on each, in parallel, without having to worry about interference/contention. (1/n)

It's been working sufficiently well for me, but there's been also many times when I waste more time than if I had just focused on the task at hand. But I guess that's the point - Im not focusing on it (because I'm actually tackling other work). (3/n)

Then, while compiling code or waiting for a result, ill check back in on the agent to see if it succeeded and/or iterate. I treat it as a low-effort-high-chaos-high-reward way to multitask more than I otherwise could. (2/n)