AI research nonprofit developing systems to discover fundamental physics
FirstPrinciples is a Toronto-based nonprofit (founded 2024) building an AI system to uncover the laws of nature through machine reasoning. The stack reveals a compute-heavy operation: HPC, InfiniBand, Slurm, Kubernetes orchestration alongside ChatGPT and Claude, indicating substantial infrastructure for large-scale model training. Hiring is decelerating (2 roles in last 30 days) despite active pain points around scaling training and attracting technical talent — typical for early-stage research orgs balancing deep talent needs with pre-revenue funding constraints.
Notable leadership hires: Director of Development
FirstPrinciples operates as a mission-driven nonprofit research organization focused on AI-assisted discovery of fundamental physics. The team (11–50 employees, Toronto-based) spans engineering, research, data, product, and fundraising functions, with a lean leadership structure. Core projects center on building and scaling their AI Physicist platform, running large-scale pre-training experiments, and executing a multi-year funding strategy. The organization runs its own compute infrastructure (HPC + cloud on AWS) rather than relying solely on commercial AI APIs, enabling long-horizon training and proprietary model development.
FirstPrinciples is developing an AI Physicist system to discover fundamental laws of nature. Active projects include large-scale pre-training experiments, LLM-based data extraction pipelines, and state-of-the-art physics methods research.
The organization uses ChatGPT and Claude for inference, paired with HPC infrastructure (Slurm, InfiniBand), Kubernetes for orchestration, and AWS for cloud compute. LaTeX, Salesforce, and Asana round out development and operations tooling.
Other companies in the same industry, closest in size