sarah guo // conviction (@saranormous) started following @wzhao_nlp on Sep 27, 2024

RT @OfirPress: Awesome idea, great execution! Its a new challenging benchmark for programming agents.

RT @asankhaya: Interesting new benchmark, the task is to recreate the Python libraries from scratch using only documentation and validated…

this is smart / cool

RT @basetenco: 🔥 We have two huge announcements today! 🔥 We’re thrilled to announce early access to our new Baseten Hybrid offering and ou…

RT @VoyageAI: 📢 Announcing a new generation of Voyage embedding models: voyage-3 and voyage-3-lite! When compared with @OpenAI's v3 large:…

The Meta Connect today was surprisingly good (sort of sad that all attention went to OpenAI, as always) - Meta Rayban glasses are underrated imo, having been using mine for a few months now, favorite part has been the seamless voice interface.

signed a new office lease and realized that the monthly rent ($4k/mo) is basically the same as renting 2 H100s monthly. not sure which is actually more valuable tbh 🙃

simplest explanations

The best part is that Matt’s so diplomatic: “for the longest time, their roadmaps sucked”

Spotify: https://t.co/RG4ulOxQer YouTube: https://t.co/tiNArOzdXF

Talked with @stanine COO @Rippling about their new launch, Talent Signal, that uses AI to understand employee performance: - looking at work product with models - initial reactions - how they considered abuse / bias - incorporation into decisions - what makes rippling powerful https://t.co/c5X69su4Mv

some feedback for candidates looking for jobs at promising startups: don’t claim “x years of experience” that’s not an outcome, that’s hanging around the hoop. you can do something in a mediocre way for years instead describe something great you have built, fixed, quota hit…

RT @KubaJaggy: Curious to see how BMW, Airbus, Boeing, Ford and four other Fortune 500 manufacturing companies already using VR/AR? We cre…

Today, experiences like this feel futuristic and amazing, but in the future, when everyone has AR glasses and MR headsets, holograms like these will be part of everyday life. This is @Arcadia_tv_, a Mixed Reality co-location fitness game. Can you imagine a future like that? https://t.co/VgnnYbHMhW

Undervalued: infectious enthusiasm

RT @basetenco: Do you know how to use TensorRT-LLM in production? If you missed our Co-Founder @defpan and @philip_kiely's talk at @aiDotE…

Before Kudos After Kudos https://t.co/M0ulZzXOkB

The @hoka Mach X 2 promises to keep 'things light, cool, and fast for next-level efforts'. But what makes the shoe so quick? In tandem with the launch of the HOKA Train At Mach Speed Strava Challenge, we explore this new 'super' shoe.https://t.co/plaFzCMXog

llm.c gang w/ @karpathy, Erik, and Arun: aka "avengers" and since Andrej's yesterday talk the "4 pandas team" lol 😂 https://t.co/e4KuXA8T7h

@karpathy on stage now talking about llm.c! https://t.co/kU27fy55l4

RT @prnvrdy: "act wisely but with @conviction" https://t.co/9LlXZl3DGB

people are sleeping on accessibility APIs in the age of AI

10/10 want to work on AI but not directly in research/engineering? Advice: work on the data engine — it’s complex, hard, creative, hugely important.

9/ We’re increasingly focused on the data engine across all our portfolio companies. The future of AI depends not just on having more data but on using it better. Smarter curation, synthetic data, high-quality human data, and robust RL will drive the next wave of capability.

1/ As AI models grow more powerful, it’s becoming clear that data strategy is just as important as model architecture for advancing capabilities. The right data—and how we use it—will determine the future of AI performance.

#airbnbxTsite https://t.co/tQoI0MbKEA

#airbnbxTsite

new batch of @conviction embed companies just landed and they’re so good 🙇‍♀️

It’s important to solve hard problems and build things that inspire a sense of awe, wonder and magic. Footages from test and integration of the pioneer N30. Volume up 🔊 🎥 @MinimalDuck https://t.co/Yt1qktPYiL

When I tried raising, a lot of VCs said it would take me $50M+ and 5-7 years to get a production boat out given the complexity and scale of the problem. That is the norm in hardware. It’s fun challenging the norm- so here I am.

https://t.co/DIDn1ayzP1

OpenAI talk at CMU today. Lines as far as the eye can see, completely filled up the largest building on campus. I waited 45 minutes and didn’t even get in. Feels like a sign of the times, the AGI revolution is here https://t.co/mP5sLY1Q9F

Yup

*exceptions, including needing compute for training FMs, which 99% of software teams shouldn’t do

forms of ego validation: - big rounds, big prices - brands, celebrities - press, conferences, klout ofc, some first-time founders understand

.@btaylor is the rare unicorn of leader, technologist, product thinker, (friend!) must-listen episode for insight into AI in the enterprise, agents in the wild, and long-term arc of tech landscape

build that datacenter baby we’re so back https://t.co/hCBPYJTD7c