Machine learning is an experimental and fast evolving discipline. The rate of business innovation is often determined by the speed of implementing, experimenting and deploying new ideas. This puts pressure on the flexibility of the developer toolchain and MLOps infrastructure.
While its roots were originally in research, the PyTorch deep learning framework has seen wide adoption across variety of industry applications in the past few years. In this talk we will cover challenges of bringing the latest AI ideas to production and the features of PyTorch that streamline this process. We will also talk about how developer-centric approach embraced by PyTorch applies to the broader MLOps ecosystem.