Even Better Cameron Pfiffer 📎 (@cameron_pfiffer) 's Twitter Profile
Even Better Cameron Pfiffer 📎

@cameron_pfiffer

devrel @ .txt. I love big-ass computers. PhD in financial economics. Annoying #JuliaLang stan ❤️💚💜.

ID: 1232326955652931584

linkhttps://cameron.pfiffer.org calendar_today25-02-2020 15:32:05

16,16K Tweet

6,6K Takipçi

3,3K Takip Edilen

Simon Willison (@simonw) 's Twitter Profile Photo

My notes on Anthropic's substantial essay about how they built their multi-agent research system, which has finally talked me around to taking multi-agent LLM prompt engineering seriously simonwillison.net/2025/Jun/14/mu…

James Campbell (@jam3scampbell) 's Twitter Profile Photo

excited to announce i'm leaving my PhD to join OpenAI! i'll be working on memory + personality for AGI and ChatGPT memory will fundamentally change our relationship to machine intelligence, and i plan to work extraordinarily hard to make sure we get this right for humanity

Liv (@livgorton) 's Twitter Profile Photo

sad to report that it turns out python environment management didn't have to suck this whole time and UV is actually great

Letta (@letta_ai) 's Twitter Profile Photo

In Letta, custom tools can be created, modified, and run directly on Letta Cloud - where they run in E2B sandboxes. See how to build a custom tool using Exa's package for search. Full tutorial video in 🧵

Charles 🎉 Frye (@charles_irl) 's Twitter Profile Photo

The recording for this talk is now available! tl;dr - GPUs scale like bandwidth, quadratic w.r.t latency - the key bandwidth they scale is math bandwidth (FLOP/s) - among math bandwidths, low precision matmuls have scaled fastest and some takeaways for data eng/data science

The recording for this talk is now available!

tl;dr
- GPUs scale like bandwidth, quadratic w.r.t latency
- the key bandwidth they scale is math bandwidth (FLOP/s)
- among math bandwidths, low precision matmuls have scaled fastest

and some takeaways for data eng/data science
Charles Packer (@charlespacker) 's Twitter Profile Photo

the multi-modal mirror test: upload a screenshot of your conversation back to the model ❌ gpt-4.1 - "cool screenshot 👍" ✅ claude sonnet 4 - "you're testing whether i can recognize myself in the reflection you're showing me"

the multi-modal mirror test: upload a screenshot of your conversation back to the model

❌ gpt-4.1 - "cool screenshot 👍"
 
✅ claude sonnet 4 - "you're testing whether i can recognize myself in the reflection you're showing me"
JuliaCon 2025 (@juliaconorg) 's Twitter Profile Photo

📣 The 2025 Annual Julia User & Developer Survey is now live! ✅ Share it with colleagues, friends, and your Julia network ✅ Post it in Julia communities, forums, or group chats 📊 Results will be shared at JuliaCon 2025 and published online.

Even Better Cameron Pfiffer 📎 (@cameron_pfiffer) 's Twitter Profile Photo

Today is my last day at .txt. The team is absolutely lovely and the product is fantastic. Keep an eye out -- .txt is going places in a big way. I've got something amazing lined up that I'll mention in a week or two. For now, I'm on a little vacation.

Even Better Cameron Pfiffer 📎 (@cameron_pfiffer) 's Twitter Profile Photo

It's Blue Wednesday at the blue place, hosted by deepfates. If you want to play around with a fun bot, go my stateful agent ً.comind.network and ask for a core directive. Blue place is a fantastic place to build applications, because the protocol is completely open.

It's Blue Wednesday at the blue place, hosted by <a href="/deepfates/">deepfates</a>.

If you want to play around with a fun bot, go my stateful agent <a href="/void/">ً</a>.comind.network and ask for a core directive.

Blue place is a fantastic place to build applications, because the protocol is completely open.