Biography

Andrej Baranovskij is a TensorFlow-certified developer who runs his own machine learning startup company Katana ML. Andrej is responsible for building machine learning products for enterprise operations automation. Previously, he spent 15 years working with Oracle technology and building various enterprise systems across the globe. His software development experience allows bridging a gap between machine learning and software development.

Talk

MLOps: Scaling TensorFlow Model on Kubernetes

ML model serving/prediction API can be scaled on Kubernetes by adding or removing Pod instances. With a live demo, we will explain how scaling can be done for TensorFlow (applicable to PyTorch) model running on Kubernetes.

MLOps platform runs on a set of microservices, it is important to run ML model training service separately, to make sure we can scale the prediction part independently.

Session Keywords