Sarah Chieng @ ICML

after a month of nights and weekends.... i've actually wanted to post on YouTube again for a really long time, but never got the courage to commit to making another video. I feel like there was so much to navigate after graduating, quitting my first job, being unemployed twice,…

Let's talk about MoE: 🔶 How many experts should you use? 🔶 How does dynamic routing actually behave in production? 🔶 How do you debug a model that won’t train? 🔶 What does 8x7B actually mean for memory and compute? 🔶 What hardware optimizations matter for sparse models?…

Super happy to be joining the Growth team with @SarahChieng and @learnwdaniel at Cerebras! Could not be joining a cooler team with cooler people. Keep an eye out for regularly shipped products and demos coming soon 👀

YC startup school 2025 documented (me sardined in a room of high schoolers) 👇👇👇 youtu.be/t5UCwcYswdE?si…

Over this last year, @amandamhuang and I started running 'Cafe Compute' SF's first, regular late-night coffeeshop for engineers, founders, writers...(anyone) to co-work or hangout late at night. We always bring in Bryan and Annie, two amazing baristas to serve free coffee and…

Free, instant access to 1M inference tok/day @CerebrasSystems is excited to launch Cerebras Supernova! Developers get 1M free inference toks/day, and if you're a startup, get up to 3 months FREE of increased context windows, rate limits, and 245M tok/day. All on the world's…

Imagine your inference provider doesn’t take down prod 💅 @GroqInc @CerebrasSystems

“Even though Diet Coke might not want to be associated with cigarettes, it’s a truth about the brand that resonates with millions of people.” Day 1 of trying new things 😵‍💫

life in silicon valley

BREAKING: MIT just completed the first brain scan study of ChatGPT users & the results are terrifying. Turns out, AI isn't making us more productive. It's making us cognitively bankrupt. Here's what 4 months of data revealed: (hint: we've been measuring productivity all wrong)

Developers automated the web. Now everyone can. Announcing Director and our $40M Series 🅱️

I am subscribed

2 hired, 2 more spots. If you're interested, DM me your resume and why you're a good fit :)

Last week, NVIDIA proudly announced their latest milestone: 1,000 tokens per second on Meta’s 400B-parameter Llama 4 Maverick, using 8 of their upcoming Blackwell GPUs on a DGX B200. This was hailed as a breakthrough—proof that NVIDIA remains the gold standard in AI inference.…