Build and optimize AI infrastructure to measure and maximize compute efficiency for training and inference workloads, ensuring accurate capacity planning and performance at scale.
10 jobs found at Physical Superintelligence
Build and optimize AI infrastructure to measure and maximize compute efficiency for training and inference workloads, ensuring accurate capacity planning and performance at scale.
Overview Physical Superintelligence (PSI) is building AI systems to discover and commercialize new physics at scale. Data centers are our first commercial market, where…
AI research role at Physical Superintelligence (psi), a startup building agents and RL training systems that learn to do physics. Day to day: train RL agents for scientific discovery, design physics-reasoning benchmarks and reward structures, scale distributed training, and publish results alongside physicists and infrastructure engineers in Boston.
Build and deploy AI-driven physics simulations and digital twins for real-world engineering problems, shipping production-ready systems that outperform traditional workflows.
Build the user-facing platform for a physics-focused AI startup: SDKs, chat interfaces, campaign tools, and partner surfaces using React, TypeScript, and Python.
Build and maintain the distributed training and inference infrastructure that powers a startup’s AI research, using PyTorch, vLLM/SGLang, and GPU schedulers to scale physics-discovery models.
Lead a team of physicists to run AI-driven physics research programs, verify AI outputs, and scale scientific discovery for a startup building physical superintelligence.
Lead a team building AI models and agents to autonomously discover and verify new physics at scale, ensuring research translates into trusted scientific breakthroughs.
Overview Physical Superintelligence is a startup with roots at Google, NVIDIA, Harvard, Meta, MIT, Oxford, Johns Hopkins, Cambridge, and the Perimeter Institute building…
Overview Physical Superintelligence is a startup with roots at Google, NVIDIA, Harvard, Meta, MIT, Oxford, Johns Hopkins, Cambridge, and the Perimeter Institute building…