Varun Singh (@vr000m) 's Twitter Profile
Varun Singh

@vr000m

@trydaily. ex-CEO @callstatsio acq’d by $eght. earlier multimedia protocols and video at $NOK, $STM, $LGE, phd @aalto. Focus on growth, revenue. 🇺🇸🇫🇮🇮🇳

ID: 7948542

calendar_today04-08-2007 05:34:22

9,9K Tweet

1,1K Followers

2,2K Following

Michelle Fang 🌁 (@michelleefang) 's Twitter Profile Photo

introducing the Starter Guide to SF — a free wiki for any founder new to or considering moving to SF. everything i wish i knew & aggregated community wisdom accumulated over the past 3 years in san francisco (esp AI communities, events, and more) startertosf.guide

introducing the Starter Guide to SF — a free wiki for any founder new to or considering moving to SF.

everything i wish i knew & aggregated community wisdom accumulated over the past 3 years in san francisco (esp AI communities, events, and more)

startertosf.guide
swyx (@swyx) 's Twitter Profile Photo

AI Engineer David Cramer alex duffy Devansh Daniel Chalef Harrison Chase Denys Linkov Dylan Patel Brooke Hopkins Brian Balfour tracks - MCP: youtube.com/watch?v=z4zXic… - Search: youtube.com/watch?v=a0TyTM… - SWE Agents: youtube.com/watch?v=U-fMsb… - Evals: youtube.com/watch?v=Vqsfn9… - Reasoning: youtube.com/watch?v=-9E9_2… - GraphRAG: youtube.com/watch?v=RR5le0… - RecSys: youtube.com/watch?v=3k4a0P… - Tiny Teams:

François Chollet (@fchollet) 's Twitter Profile Photo

Beyond the perhaps superficial semantic distinction between "reasoning" and "pattern matching", there is a fundamental gap in the practical capabilities and behavior of these systems. You don't create an invention machine by iterating on an automation machine.

Vaibhav (VB) Srivastav (@reach_vb) 's Twitter Profile Photo

Let's goooo! kyutai just dropped SoTA Speech to Text transcriptions model - CC-BY-4.0 Licensed 🔥 > kyutai/stt-1b-en_fr (1B params, 500ms delay, English & French) > kyutai/stt-2.6b-en (2.6B params, 2.5s delay, English-only, higher accuracy) > Capable of 400 real-time

Let's goooo! <a href="/kyutai_labs/">kyutai</a> just dropped SoTA Speech to Text transcriptions model - CC-BY-4.0 Licensed 🔥

&gt; kyutai/stt-1b-en_fr (1B params, 500ms delay, English &amp; French)
&gt; kyutai/stt-2.6b-en (2.6B params, 2.5s delay, English-only, higher accuracy)

&gt; Capable of 400 real-time
Vaibhav (VB) Srivastav (@reach_vb) 's Twitter Profile Photo

Pretty Insane - SoTA Text to Speech model capable of English AND Hindi - 3B Llama backbone - Apache 2.0 licensed 🔥 > Sub 80 ms latency > Supports both English, Hindi including code-mix > Runs in a free google colab too 🤯 Best part: They're actively working on other languages

kwindla (@kwindla) 's Twitter Profile Photo

swyx AI News by Smol AI dex As with all great ideas, I'm sure there could have been parallel creation. But I first encountered the "context engineering" phrase in Dex's 12-factor-agents.

<a href="/swyx/">swyx</a> <a href="/Smol_AI/">AI News by Smol AI</a> <a href="/dexhorthy/">dex</a> As with all great ideas, I'm sure there could have been parallel creation. But I first encountered the "context engineering" phrase in Dex's 12-factor-agents.
Daniel Green (@dgrreen) 's Twitter Profile Photo

🎙️🤖 Jon Taylor & kwindla Daily just released the Pipecat AI Voice UI Kit – React components, hooks, and templates to speed up shipping voice-first apps. Could this be one of the building blocks of Voice AI’s “ChatGPT moment”?

Vaibhav (VB) Srivastav (@reach_vb) 's Twitter Profile Photo

Google COOKED yet again - Multimodal Gemma3n 4B and 2B now available in Transformers, vLLM, MLX AND Llama.cpp 🤯 The model can see, hear and type - all in 140 languages ⚡ Best part: You can fine-tune it in a FREE google colab 🤗 Enjoy!

Google COOKED yet again - Multimodal Gemma3n 4B and 2B now available in Transformers, vLLM, MLX AND Llama.cpp 🤯

The model can see, hear and type - all in 140 languages ⚡

Best part: You can fine-tune it in a FREE google colab 🤗

Enjoy!
kwindla (@kwindla) 's Twitter Profile Photo

Hot off the presses: an open source Voice AI UI kit that you can use with any model, API, or agent framework. Beautiful React and JavaScript components for voice agents. The voice-ui-kit supports any network transport and any server-side code/API in the Pipecat ecosystem:

Dylan Field (@zoink) 's Twitter Profile Photo

Sharing an update on Figma: we publicly filed our S-1 with the SEC today, and have applied to list on the New York Stock Exchange under the symbol “FIG.” figma.com/blog/s1-public

Simon Willison (@simonw) 's Twitter Profile Photo

I figured out how to add the official Playwright browser automation MCP to Claude Code. Run this before you start "claude": claude mcp add playwright npx '@playwright/mcp@latest' Now Claude Code can use a Chrome browser directly! Here's my TIL: til.simonwillison.net/claude-code/pl…

Gabriele Berton (@gabriberton) 's Twitter Profile Photo

Another issue is that different hardware might produce slightly different logits, which is enough to change the order of two tokens and completely break the de-compression But overall, nice paper and interesting idea [6/6] arxiv.org/abs/2306.04050

kyutai (@kyutai_labs) 's Twitter Profile Photo

Kyutai TTS and Unmute are now open source! The text-to-speech is natural, customizable, and fast: it can serve 32 users with a 350ms latency on a single L40S. Try it out and get started on the project page: kyutai.org/next/tts

Vaibhav (VB) Srivastav (@reach_vb) 's Twitter Profile Photo

Kyutai released their Streaming Text to Speech model, ~2B param model, ultra low latency (220ms), CC-BY-4.0 license 🔥 Trained on 2.5 Million Hours of audio, it can serve up to 32 users w/ less than 350ms latency on a SINGLE L40 🤯 Incredible release by kyutai folks, go check

Varun Singh (@vr000m) 's Twitter Profile Photo

Awesome to see Michael and team’s perseverance pay off they have been diligently working on making serverless ai a reality! Congratulations