At Sciloop, we're trying to automate scientific research and discovery to push science to new frontiers. Starting with ML research, we're building tools that let researchers focus on ideas while we handle the experimentation and writing code. We're designing tools we wish existed—autonomous, flexible tools that make it easier to do better research quickly and in a more reproducible manner.

The best models in the world still can't reliably solve an Olympiad problem or reason through open research questions. We work with LLM companies to change that—supplying the hardest problems humans can write, sourced from competitive mathematics and the bleeding edge of scientific research. The kind of problems that actually push a model forward. Learn more →