Co-Founder and Chief Science Officer at @andTruthPBC, Associate Professor at @NUSComputing. Proving the world's software correct, one function at a time.
I'm thrilled to bring &Truth out of stealth as its co-founder and Chief Science Officer. With a team I'm proud to be part of, we're setting out to specify and prove all the world's software correct and secure.
If you geek out over code people can understand and trust, join us!
We are out of stealth! As AI writes more code, we're building tools to prove programs meet precise requirements, with Lean proofs anyone can independently check. Let's make verified software the norm in every language and application.
Come take a look: https://t.co/NUGQLplyrZ
Our &Atlas draws code as "constellations": dots are functions, circles are file clusters. Deep call chains at the center, leaves at the rim. Green = verified.
Which famous project is this? Hint: there are several copies in your pocket.
Check your guess: https://t.co/grPzXcLczY
&Truth co-CEO @ClarkBarrett7 argues that software can be made unhackable by mathematically proving that entire classes of bugs cannot exist:
"We've been conditioned to believe that any system can be hacked."
"But the truth is that a hack exploits a bug, and if you can mathematically prove that a certain class of bugs no longer exists, then that system actually becomes unhackable."
"There have been some success stories in the history of formal methods where a system was built that could not be hacked."
"In the past, it's been very expensive to do that, but that’s also changed now."
@andTruthPBC
What do the recent agentic verification results mean for formal methods research? Read my thoughts on the PL Perspectives blog.
P.S. I'm recruiting PhD students!
https://t.co/Jek5gh1Fq7
@HerrDreyer You might not have noticed, but it’s much the same in academic publishing: everyone checks out and cites new papers by celebrities known for their eloquent writing, while ignoring work by lesser-known authors is pretty much okay—unless those authors happen to be your reviewers.
Very much agree with this: academic publications are (a) an okay way to get expert feedback and (b) a terrible way to disseminate ideas. That's why I've started blogging:
https://t.co/xI3PoV00Js
I am increasingly of the opinion that the academic paper is no longer a good medium for disseminating CS research. I think we should move to (1) blog posts and (2) books.
Thank you, @ilyasergey for all the guidance and support along the years! It’s been an amazing time. And thank you to everyone who has been along on this journey with me - I am immensely grateful. Thank you and I can’t wait for what comes next!
Huge congratulations to my PhD student @GeorgePirlea, who just defended his thesis! George switched my whole lab to Lean, masterminded https://t.co/G5vmNftpT0, and got our multi-modal verification-as-a-library agenda off the ground. Very proud of him. All the best, Dr. Pîrlea!
Immensely proud of the team's achievement! Today @andTruthPBC launched &Atlas, the world's first interactive map tracking progress toward safe, mathematically verified software.
The journey towards verifying all the code starts with a single spec, but at least now we have a map.
1/5
Today we’re launching &Atlas: a map of 100M+ lines of open-source C across 1,000+ codebases, tracking progress toward provably verified software.
The scale and challenge are enormous. Explore this universe: https://t.co/VHMHHgJGnD
1/5
Today we’re launching &Atlas: a map of 100M+ lines of open-source C across 1,000+ codebases, tracking progress toward provably verified software.
The scale and challenge are enormous. Explore this universe: https://t.co/VHMHHgJGnD
@awsTO It's real thanks to the work of so many people across decades at this point. Formal verification is scaling and will be ready for broad adoption very soon.
And if we have a say in how this plays out, it won't be gate-kept, but open source and freely extensible and composable.
Veil 2.0 is out: verify all your distributed protocols in Lean 4.
New: verified model checkers for safety (2–20x faster than TLC) and liveness, LTL proof mode, TLA+ → Veil LLM skill, PlusCal-style DSL.
Great work by @zqy1018 and external contributors!
https://t.co/9OMsf1Kxzt
@codeofchange@angerman@zqy1018 You can run your models natively with Lean compiler: this is what we do for model checking. Certified extraction to Rust is WIP. Curious about your workflow with TLA+/Rust.