We are building software that makes machine learning execution simple and efficient. What does that mean? Together with our community, we engineer very sparse NLP and CV models that are more efficient and performant in production. They recover to baseline accuracy and you can apply your data using only a few lines of code. Why does this mater? Sparse models are more flexible and can be deployed at extreme speeds without GPUs. By using DeepSparse, our sparsity-aware inference engine, you can achieve unrivaled latency and throughout performance using commodity CPUs you already own. Check us out on GitHub and join the Deep Sparse Slack community to help us accelerate our vision of software-delivered AI.