Phaidra
AI agents for industrial control systems, specializing in power, cooling, and workload management for AI factories and data centers.
About
Phaidra builds AI agents that run the power, cooling, and workload systems behind AI factories and data centers. The pitch is "maximize tokens per watt." Two products so far: Phaidra Prism, an LLM for data center operations that helps technicians triage issues, and Phaidra Factory, agents that monitor and optimize infrastructure at scale. In an NVIDIA test, the liquid-cooling agent cut thermal spikes from AI workloads by roughly 80%.
The team's track record is in the same problem space. Members previously worked on DeepMind's AlphaGo and on the Google project that reduced data center cooling energy by ~40%. Backed by NVIDIA.
Based in the USA but 100% remote with no physical office. Team spread across the USA, Canada, Europe, and India. Documentation-first, async by default, openly modeled on GitLab's remote playbook. Unlimited PTO with a 20-day minimum, monthly remote stipend, home office allowance, company MacBook.