Victor Sanh

Founder @ Something new | AI that moves things | Prev. early crew @huggingface 🤗 | AI x NYC

[Victor Sanh @SanhEstPasMoi]

@ClementDelangue gg, let's gooooo!

[Victor Sanh @SanhEstPasMoi]

It’s so fascinating to me that many of the researchers working at the frontier of AI, arguably the most consequential technology of this decade, are quietly very depressed and anxious. The only thing keeping them going is FOMO, not fun or passion.

[Victor Sanh @SanhEstPasMoi]

@fluorane @appliedcompute @trajectorylabs @EngramLab hedging the pilot programs

[Victor Sanh @SanhEstPasMoi]

Here's a detected bug, is it legit? If so, please fix.
```
{INSERT_BUG_DESCRIPTION}
```

[Victor Sanh @SanhEstPasMoi]

https://t.co/LtxIkJELeV

[John Sarihan @jsarihan]

Today, we're launching Crosby Intelligence to push the frontier of legal AI forward.

[Victor Sanh @SanhEstPasMoi]

no better time to build sovereign ai

[Victor Sanh @SanhEstPasMoi]

@Polymarket that's the way it should be

[Victor Sanh @SanhEstPasMoi]

@natolambert @allen_ai Legend!

[Victor Sanh @SanhEstPasMoi]

would that work with nyc cockroaches?

[Bull Theory @BullTheoryio]

BREAKING: Google is planning to release 32 million mosquitoes across Florida and California.

The company has asked the EPA for permission to proceed, with the public given until June 5 to respond.

The mosquitoes are infected with Wolbachia bacteria, which stops them from reproducing and slowly collapses the wild population from within.

Google's previous Debug Project trial in California's Central Valley nearly eliminated mosquitoes from three test sites entirely. A separate trial in Singapore cut dengue cases by 70% within 12 months.

Google has now released over 1 billion mosquitoes across four continents. This new proposal is the largest deployment in US history.

[Dan Constantini @danoandco]

Introducing https://t.co/ymuN68eFXG Dev Station: every workspace now has a persistent, multi-repo dev station that tasks fork from.

@twill_ai station is now the source of truth for the project: repos connected, dependencies installed, env vars configured, dev/test servers running, and project context warm.

Agents and humans can both configure it, so the workspace keeps getting closer to the way your team actually builds software.

Tasks are no longer independent cold starts. Each task forks from the station’s current working environment into its own isolated sandbox. Powered by @daytonaio.

That means:
• No cold-start, so tasks start/finish faster.
• Multi-repo work across frontend, API, and shared libraries.
• Less setup work for agents, which means fewer tokens spent re-setting the dev env.

This is the model we think coding agents need: a persistent workspace that stays warm, then forks cleanly for each job.

Try it at https://t.co/BjDTPcquKi

[Dar Mehta @dar_mehta]

Hosting a happy hour next Thursday at @CVPR with @iScienceLuvr from @SophontAI!

Also will be talking about our bet on causal intelligence at CV4Science workshop.

[Vik Paruchuri @VikParuchuri]

Announcing Surya OCR 2:

- 650M params
- 83.3% olmocr bench score (top under 3B)
- 87% on internal 91-lang benchmark
- 5 pages/s on RTX 5090
- Runs on CPU, GPU, MPS

[Victor Sanh @SanhEstPasMoi]

@ShayneRedford @alex_pentland @sarahookr @PeterHndrsn congrats!!

[Victor Sanh @SanhEstPasMoi]

@Majdaxxx fun!!

[Victor Sanh @SanhEstPasMoi]

your average nyc party chitchat:
- favorite pizza spot
- favorite bagel spot
- favorite tacos spot
- which neighborhood they “discovered”
- if expat, what visa they are on
- rationalizing {insert expensive private club} membership
- rent
- mamdani

[Victor Sanh @SanhEstPasMoi]

RL is hard.
Have you tried double RL?

[Victor Sanh @SanhEstPasMoi]

@mfishbein they built a product that addresses their customer's problem.

my point is independent of which sdk the solution is built on top of

[Victor Sanh @SanhEstPasMoi]

the reality of the vast majority of vertical ai app layer companies:

website describes one single agent that works across all use-cases listed, across all customers

FDEs have built N+ agents/systems, where N is the number of customer. some common abstractions but maintenance and improvement is largely a pain in the butt. each new customer is increasing complexity with small and odd quirks.