🗼

Explore roles, hiring trends, and funding at AI companies on breakout trajectories. Updated daily.

Inferact

Inferact is the commercial company built by the creators and core maintainers of vLLM — the most widely deployed open-source engine for running large language models efficiently. vLLM was originally developed at UC Berkeley's Sky Computing Lab and now runs on 400,000+ GPUs worldwide, with 2,000+ open-source contributors. Inferact operates an "open core" model: vLLM stays community-owned and PyTorch Foundation-governed, while Inferact sells managed services, optimized cloud deployments, and enterprise-grade tooling on top. The core technology uses techniques like PagedAttention to maximize GPU throughput and minimize memory waste, supporting NVIDIA, AMD, and Google TPU hardware. Buyers are enterprises and cloud providers that need to serve LLMs — including diffusion models, MoE architectures, and agentic workloads — at low cost and high reliability. The founding team (CEO Simon Mo, CTO Woosuk Kwon, Chief Scientist Kaichao You, and co-founder Roger Wang) has advisors Ion Stoica and Joseph Gonzalez of UC Berkeley, giving it unmatched open-source credibility and upstream control.

$800M post-money valuation · 14 employees · 32 open roles

Website · Careers

Inferact

Open roles