←
Applied Compute334 posts
Applied Compute avatar✓

Applied Compute ✓

@appliedcompute

The best AI is built, not bought. Our platform, Applied Compute Agent Cloud, is now in private beta. Book a demo below.

◷ Joined July 2012▧ 107 Photos and videos
Applied Compute avatar
Applied Compute@appliedcompute·Nov 2

Every company should own the AI it runs on. Applied Compute exists to make that possible. We’ve crossed $100M in annualized run rate and raised $350M at a $3.25B valuation, led by @eladgil. 40% of customers now use Applied Compute for both training and inference.

Post attachment
Yash Patil avatar
⟳ Applied Compute retweeted
Yash Patil@ypatil125·Nov 12

Applied Compute has acquired Delegate! We were one of Delegate’s first customers, and their product is how we kept up with hundreds of accounts before we had a sales team. Felix Su and Jason Yang’s attention to detail, pace of shipping, and product sense are unparalleled, and I’ve wanted to work with them ever since we met. Before Delegate, they helped build Scale AI’s enterprise business from the ground up, getting AI into production at some of the largest companies in the world. Much of our team is former founders like Felix and Jason. They bring a rare combination of ambition, agency, and the ability to run through walls. Welcome, Felix and Jason! I couldn’t be more excited to build with you. 🚀 appliedcompute.com/company/d…

Post attachment
Applied Compute avatar
Applied Compute@appliedcompute·Nov 11

Poker is a hard post-training problem. Many players, hidden information, and results only show up at the end of a long game. It's even harder with only 24 hours on the clock. We're sponsoring the PokerBench track at the @ValsAI hackathon this weekend to see what teams can train on AC2. Spots are limited. We'll see you at the table.

Post attachment
Sequoia Capital avatar
⟳ Applied Compute retweeted
Sequoia Capital@sequoia·Nov 10

Want to come to the No Limit Finals in San Francisco? - Top 20 student teams across 800+ participants build pokerbots to compete for a $100K grand prize. - @OpenAI’s @polynoamial joins us for a talk with @Alfred_Lin. - Founders and top players from @Etched, @Factory, @appliedcompute, @DoorDash, @fal, @Profound, @getserval, @sailresearchco, @conwayinc and more compete to determine the best poker player in AI. Proceeds go to @khanacademy. There are still a few spots left for partner companies to compete. - We also have a few guest list spots available for spectators. Interested? Email pokerbot@sequoiacap.com In partnership with @citsecurities and @OpenAI

Post attachment
Abhishek Malani avatar
⟳ Applied Compute retweeted
Abhishek Malani@abhishekm1636·Nov 10

Who’s the best poker player in AI? We’re about to find out. The No Limit Charity Cup is Sunday, 11/15 in SF. $50K per seat. Founders and top players from @Etched, @Factory, @appliedcompute, @DoorDash, @fal, @Profound, @getserval, @sailresearchco, @conwayinc and more go head to head, with proceeds going to @khanacademy. Think you’ve got what it takes? We still have a few seats for partner companies to send a delegate to compete. Want to come watch? DM me. Bragging rights on the line. Thanks to our partners: @citsecurities @OpenAI.

Post attachment
Applied Compute avatar
Applied Compute@appliedcompute·Nov 9

We incorporated quantization-aware distillation (QAD) into our training platform, AC2. This lets enterprises run their custom models on less hardware without giving up the accuracy they trained for: 1) Compressing the right parts of a model cut its memory by about a third without hurting performance. 2) QAD can win back the accuracy that compression usually costs, in some cases matching the original model.

0:23
Sam Denton avatar
⟳ Applied Compute retweeted
Sam Denton@samueldenton·Nov 6

This is it. If I were in charge of a company, it'd be my number 1 priority to get this working ASAP. No RL envs, no simulated tasks, no large efforts for data creation. Just straight production traffic -> weights.

0:10
Raymond Feng avatar
⟳ Applied Compute retweeted
Raymond Feng@raymondmfeng·Nov 6

maybe the real training data was the work we did along the way

0:10
Applied Compute avatar
Applied Compute@appliedcompute·Nov 6

The rest of the mismatch comes from quantization, engine differences, and staleness; we use score centering to mitigate these off-policy effects. It learns faster and higher than importance sampling, works with real top-p and top-k settings, and stays stable even when run asynchronously for faster rounds. A sampled KL penalty keeps run variance low.

Post attachment
Applied Compute avatar
Applied Compute@appliedcompute·Nov 6

Prod RL has to stay stable across every redeployment, since the end-to-end prod RL run is expensive and rolling back a few checkpoints may not rescue a collapsed run. Most of the instability comes from mismatch between training and test time, which starts with formatting the data. We record the exact tokens the deployment generated to avoid mismatch from inconsistent tokenization.

Post attachment
Applied Compute avatar
Applied Compute@appliedcompute·Nov 6

Production traffic is full of real experience from economically valuable work, but very little of it currently makes it back into training. We built a recipe for production RL that steadily improves a model straight from its deployment’s data stream, then redeploys it to keep learning. We are sharing some of our initial learnings from a recipe capable of continual learning in production.

0:10
Rahil Verma avatar
⟳ Applied Compute retweeted
Rahil Verma@RahilVerma13·Nov 4

I built a bot that reviews code like me, then had Ari, our agent on AC2, autoresearch an eval for it and hill-climb it overnight. Bug bots like Codex are great for coverage, but they don’t know which shortcuts I’ll let slide, or that I review every PR asking “would we build it this way from scratch?” It just doesn’t feel like I’m getting a review from someone that understands our codebase or team. Code review has become Applied Compute’s biggest bottleneck as we write code faster. @felixs8696 joked I should build a bot since I was so behind on his PRs, so I did. I turned 3,500 of my comments across 2,000 PRs into an agent that talks and thinks like me. I’d like to think I am more complicated than an agent, and on the first pass it seemed I was right. The first version was noisy, leaving a pile of comments where I’d leave one or two - I couldn’t trust it. So I asked Ari, our agent on AC2, to build an eval and hill-climb it, prioritizing precision so the bot wouldn’t flood PRs with comments I’d never make. I turned on autoresearch overnight, and Ari: - Built a dataset of my past reviews, checking out each PR exactly as it was when I reviewed it - Built graders that verify every finding against the real code and my actual comments - Stress-tested those graders on 50 adversarial cases before trusting them - Ran ~100 experiments across prompts, models and harness changes, confirmed on a held-out set By morning, reviews were ~11 points more precise and used 22% fewer tokens. The biggest win came from how the bot explores the codebase, not from the prompt. Now it’s my first pass on my own PRs, and I trust it because it was measured on my own data. If you want to build your own evals on AC2, reach out! Checkout the autoresearch blog post here: appliedcompute.com/research/…

Post attachmentPost attachmentPost attachmentPost attachment
Yash Patil avatar
⟳ Applied Compute retweeted
Yash Patil@ypatil125·Nov 4

Inference isn't everything. We’re building the full platform for open models: training, inference, routing, synthetic data generation, and endpoints that continuously improve. Ari, our applied research engineering agent, is built directly into AC2 to automate post-training. It curates data, runs experiments, evaluates results, and iterates using the same tools our researchers use.

0:10