James Alcorn

partner at lightspeed 🇦🇺 🇺🇸

[James Alcorn @JamesAlcorn94]

reward is eval, eval is reward, reward is enough

[James Alcorn @JamesAlcorn94]

quality @kokoxsu @carloagostinel2

[koko 𝑥𝑠 @kokoxsu]

We @novaholdings crafted the most complete library of founder biographies — origin stories from the formative years of the out-of-distribution individuals behind history’s defining companies.

Founder Profiles, by Nova Global

[koko 𝑥𝑠 @kokoxsu]

We @novaholdings crafted the most complete library of founder biographies — origin stories from the formative years of the out-of-distribution individuals behind history’s defining companies.

Founder Profiles, by Nova Global

[James Alcorn @JamesAlcorn94]

@KateClarkTweets when i grow up i want to be like @EverettRandle

[Simon Mo @simon_mo_]

🦀 rustifying vLLM, one part at a time, great work @BugenZhao!

[James Alcorn @JamesAlcorn94]

@maddiehfaulkner @matanSF @FactoryAI @NEA great move @maddiehfaulkner and congrats @matanSF, @EnoReyes!

[James Alcorn @JamesAlcorn94]

@mcutler Congrats @mcutler !!

[James Alcorn @JamesAlcorn94]

@danielhanchen @UnslothAI cc @joespeez 🫡

[James Alcorn @JamesAlcorn94]

congrats to @danielhanchen, his brother / cofounder mike chen, and the entire @UnslothAI crew for joining an exclusive pytorch club. you are in good company alongside transformers (hugging face), vLLM (inferact), SGLang (radixark). proud of you fellas!

[Unsloth AI @UnslothAI]

We’re excited to share that Unsloth has joined the PyTorch Ecosystem!

Unsloth is an open-source project that makes training & running models faster, more accurate with less compute. We want AI to be accessible to all

Blog: https://t.co/viXIukVHnU
GitHub: https://t.co/aZWYAtakBP

[James Alcorn @JamesAlcorn94]

@danielhanchen @UnslothAI unsloth will remain as an independent open-source project, separate from the PyTorch Foundation

[James Alcorn @JamesAlcorn94]

training ai scientists on hypothesis → experiment → falsification is a bonafide sick bootstrap. No dramas. All class.

less obvious: are we teaching models to do science, or merely to perform science as recognized by Homo sapiens sapiens?

If a model found a genuinely alien discovery procedure, would we reinforce it - or punish it for being off-distribution from our own?

Q: Could it be superior to ours?
A: No, because it looks wrong.
To be fair, very on-brand for science. Ask Galileo.

[James Alcorn @JamesAlcorn94]

nothing new under the sun 🤷‍♂️

[James Alcorn @JamesAlcorn94]

@ml_angelopoulos 🙃

[James Alcorn @JamesAlcorn94]

@steph_palazzolo 🙌🙌

[Andrew Li @andrew_li03]

first partner we’ve met and has led us since. always fun to chat and think with about the future of AI

[Alex Shan @alexshander03]

@JamesAlcorn94 is a special person and an even more special investor.

He was one of the first to deeply understand what our technical goals were, and to believe that they were possible. When we started @JudgmentLabs, the concept that long-horizon agents required their own improvement infrastructure was radical. James has a consistent track record of finding technical teams and betting early - whether it be us, @thinkymachines , @UnslothAI , or @periodiclabs .

Partnering w/ James and LSVP together at the seed was a tough but well warranted decision. By the time our next round came, it was a foregone conclusion from both ends that we would continue to win together. I walked in with no deck and no memo, but the whole team came to consensus within a few minutes.

I am proud to partner up with James and Lightspeed as a whole as they double down into our Series A! 🚀

[James Alcorn @JamesAlcorn94]

@JudgmentLabs is building the canonical infrastructure for agents. A closed loop system to instrument, evaluate, and optimize.

Cofounders @alexshander03 @andrew_li03 and @JSCamyre understand uniquely what the eval represents - so much more than a post-hoc observability metric, it is precisely what defines agent capability and reliability.

This is a special team that got its start over a decade ago as childhood friends, and today as cofounders of this remarkable company. Alex, Andrew, and Joe have been assisted heroically by @carloagostinel2 since day 0.

@JudgmentLabs is one of our fastest growing portfolio companies and we're today thrilled to announce their Lightspeed-led seed & co-led series a, alongside friends @Greenoaks and @novaholdings.

[Alex Shan @alexshander03]

We’re launching @JudgmentLabs today and announcing $32M in funding.

As AI agents take on more of the work that creates economic value, they generate massive amounts of production data: the clearest record of how they behave with users, software, and the real world.

Judgment builds infrastructure for improving AI agents from production data.

[Lightspeed @lightspeedvp]

Three childhood best friends, each inspired by AI research, have come together to strengthen the quality layer for AI agents.

Co-founders @alexshander03, @andrew_li03, and @JSCamyre have known each other since they were kids, when they bonded over a shared excitement for software engineering and artificial intelligence. Now they’ve teamed up to bring their experience to the AI agent space with @JudgmentLabs.

Judgment Labs aims to quality-check and continuously improve agents with every interaction. We’re excited to announce that we partnered with Judgment Labs on their combined $32M in funding for their Seed and Series A rounds.

Agents are a leading factor in the current AI conversation. We believe systems built to optimize this technology are positioned to stay ahead of the pack.

[James Alcorn @JamesAlcorn94]

@alexshander03 @JudgmentLabs lfg

[Alex Shan @alexshander03]

We are honored to partner with you all!

@JamesAlcorn94 @guruchahal