Strange Loop
Strange Loop
Four AI researchers who just graduated from Princeton, with work spanning reinforcement learning, LLM pre- and post-training, and agentic systems shipped to enterprises and governments.
Four AI researchers who just graduated from Princeton, with work spanning reinforcement learning, LLM pre- and post-training, and agentic systems shipped to enterprises and governments.
Meet
Meet
Rhiaan, Shlok, Chirayu & Tharun
Rhiaan, Shlok, Chirayu & Tharun
Building environments that teaches AI how to turn questions into discoveries.
Building environments that teaches AI how to turn questions into discoveries.
Our north star is artificial intelligence that makes scientific discoveries independently and far faster than humans. We believe that reaching this goal begins with rich data showing how researchers solve real, open-ended problems.
Strange Loop brings models, compute, experiments, data, logs, knowledge, and collaborators into one research workspace. It accelerates researchers’ work while capturing how their ideas and decisions evolve from an open question to an evidence-backed conclusion. With less than a week of research on our platform, we published two ICML 2026 workshop papers in robotics and LLM pre-training, and more than 200 researchers at ICML joined our waitlist.
We turn these real research trajectories into rigorously reviewed reinforcement-learning environments. These environments teach foundation models to run experiments and reach meaningful scientific conclusions, starting with academic machine-learning research. We're in early talks with OpenAI, DeepMind, and xAI to license this data.
Our north star is artificial intelligence that makes scientific discoveries independently and far faster than humans. We believe that reaching this goal begins with rich data showing how researchers solve real, open-ended problems.
Strange Loop brings models, compute, experiments, data, logs, knowledge, and collaborators into one research workspace. It accelerates researchers’ work while capturing how their ideas and decisions evolve from an open question to an evidence-backed conclusion. With less than a week of research on our platform, we published two ICML 2026 workshop papers in robotics and LLM pre-training, and more than 200 researchers at ICML joined our waitlist.
We turn these real research trajectories into rigorously reviewed reinforcement-learning environments. These environments teach foundation models to run experiments and reach meaningful scientific conclusions, starting with academic machine-learning research. We're in early talks with OpenAI, DeepMind, and xAI to license this data.