Applied Compute builds Specific Intelligence for enterprises by training custom models and deploying in-house agent workforces to unlock latent company knowledge. The company has raised $80 million and counts Fortune 50 companies among its clients, including DoorDash, Cognition, and Mercor.
The firm's technical focus spans reinforcement learning, post-training techniques, enterprise AI deployment, and machine learning systems infrastructure. It builds training stacks, agent platforms, and tools entirely in-house for clients, enabling rapid iteration - models can be validated and deployed in days rather than months.
The team is composed heavily of former founders, alongside top AI researchers and International Math Olympiad winners. Founding members bring experience from OpenAI, including work on the agentic Codex project, contributions to the first RL-trained reasoning model (o1), and building ML systems infrastructure for reinforcement learning training. Engineers embed directly within client teams as part of the deployment model.
