Nonprofit AI research institute developing open foundation models and reasoning systems
AI2 is a Seattle-based nonprofit research institute founded in 2014 that develops foundational AI models, reasoning systems, and robotics. The tech stack—Python, Go, NCCL, InfiniBand, plus enterprise finance tools (Workday, Sage Intacct)—reflects a research-heavy organization scaling compute infrastructure. Hiring is almost entirely research-focused (17 of 25 active roles), with senior researchers leading projects on language model pre-training, agentic behavior benchmarking, and the OLMo training pipeline. Internal pain points center on grant management, fund tracking, and technical delivery acceleration—typical friction points for research nonprofits managing restricted funding and cross-team dependencies.
AI2 conducts nonprofit research and development in artificial intelligence, focusing on foundational models, natural language processing, computer vision, and machine reasoning. The organization publishes open-source models and datasets alongside robotics and conservation initiatives. Operating at 201–500 employees with headquarters in Seattle, AI2 is grant-funded and runs a research-driven operational model. Current work spans language model pre-training (including the OLMo project), agentic model evaluation, benchmark design, and infrastructure for large-scale ML pipelines. The institute also operates Beaker, an internal scheduling platform for research compute.
Core research stack: Python, Go, NCCL, InfiniBand. Enterprise operations: Workday, Sage Intacct, Adaptive Planning, ADP, UKG Ready. Design and collaboration: Figma. Retrieval-augmented generation (RAG) and internal scheduling via Beaker platform.
Language model pre-training and the OLMo project; agentic model training and evaluation; benchmark and environment design for reasoning; OLMo infrastructure and training pipelines; Beaker scheduling platform; and scientific synthesis applications.
Other companies in the same industry, closest in size