PostDoc position available in my Formally Verified Security group at MPI-SP. Looking for excellent research track record in PL (POPL 2024 and ICFP Conference) and/or security (@acm_ccs and IEEE CSF ).
Details here: groups.google.com/g/fstar-mailin…
#academicjobs #research #security

I just published a bit on GitHub Artifact Attestations and how they relate to SLSA and Sigstore. Hopefully it can serve as a helpful supplement to GitHub's own blog post.
ianlewis.org/en/understandi…

I can’t get over how elegantly autodiff can be implemented in Haskell with dual numbers via typeclasses.

If you want to see how to build a tiny autodiff lib + a working feedforward net on top of it, all in pure Haskell, from scratch, I just released this: github.com/milton-l/micro…

'Mathematica, a Secret World of Intuition and Curiosity' is out tomorrow — phew!
It took me 20 years to fail to write it and 1 to actually do it.
Thanks Steven Strogatz, Terry Tao, Hugo Duminil, Dr Eugenia Cheng --no advice please--, Ben Orlin, Ian Stewart for the great blurbs!
yalebooks.yale.edu/book/978030027…

Don't let AI risks slow you down 🏃🏼‍♀️💨

Join Google Cloud Security experts for a webinar and discover the best practices for safeguarding your AI environment.

Register now: bit.ly/3wZM7eE

#GenAI #AISecurity #Cybersecurity

To not know maths is a severe limitation to understanding the world!

Sharing the slides from my keynote 'Democratizing Fuzzing at Scale' at the National University of Singapore: Fuzzing and Software Security Summer School 2024 - drive.google.com/file/d/1lUFIug…

Classic Jupyter notebook fans, I've great news: nbclassic, the original (and still my fave 😁) notebook just got a new release -- the 1st in >1 year!

Answer.AI partially sponsored this release, in order to get a years-old notebook-trust bug fixed.

Nbclassic still matters!🧵

Want database diagrams, table/view/column autocomplete, and other goodies when using sqlite in Project Jupyter?

Then you might be interested in my new project `fastlite`, which adds some cool stuff to Simon Willison's marvellous sqlite-utils project.

Link in next tweet.

Amazing thread and the associated writeup is even better:
hazyresearch.stanford.edu/blog/2024-05-1…

'And we ask: if your matrix multiply is smaller than 16x16, are you sure what you’re doing is AI?' 😂

Microsoft: “We’re putting security first.”
Also Microsoft: “Windows now comes with what’s functionally a screen capturing keylogger.”
🤡🤡🤡

There are 2 mistakes you can make about LLMs:

① Thinking everything LLMs say is correct, they can reason, and with a bit more scale they’ll get us to superintelligence

② Thinking LLMs are good for almost nothing—they are FAR better at all #NLProc tasks than previous methods

Yann LeCun Elon Musk Regulating Open Source community and smaller companies is the worst thing that can happen to AI.

1. I'm legit shocked by the design of Meta's new notification informing us they want to use the content we post to train their AI models. It's intentionally designed to be highly awkward in order to minimise the number of users who will object to it. Let me break it down.

We just created a blog for OSS-Fuzz!

The first post is about extending our LLM fuzz harness generation work to completely new, unfuzzed projects: blog.oss-fuzz.com/posts/introduc…

One reason why I stick with vim instead of modern IDEs that have a bespoke project organization and cannot handle if you deviate from it

Fall of Civilizations is hands down the best podcast on history that I've ever listened to. I definitely recommend the book for all history nerds out there.

I’ve been very much enjoying this talk series given by Ana Caraiani at IAS, which truly does justice to its title: a glimpse into the Langlands Program.

youtu.be/2bKeI7GulVE?si…

surely this is the 3-day weekend that will fix my crippling burnout