We're a foundation model company focused on math evaluations and datasets, while building cutting-edge avatar AI and real-time models for the next generation of AI systems.
We believe the most promising path to robust AI systems is through rigorous evaluation and expert-curated datasets. Our approach combines advanced mathematical reasoning tests, real-time AI model development, and foundation model training to build more capable and reliable AI systems.
We specialize in two core areas:
Our team brings experience from UWaterloo, Reka AI, Hedra AI, and ZJU, with deep expertise in building video and real-time models. We are a small group of engineers and researchers working to solve fundamental problems in AI evaluation and model development. If this sounds interesting, we would love to hear from you.