Most of the time for Data scientists, Machine Learning Models are relatively easier to build but they are very hard to roll out. Data scientists usually understand their tools and build using their local systems or cluster, but they don’t know how to roll them out to the production. Along with this, scalability in real time on production environments is challenging.
And then on top of it, every cloud provider has its own way of handling MLOPs. AWS has SageMaker, Google has AI Platform, and Azure has ML Studio.
What options are available for something that can run across cloud providers? Ever thought of using machine learning in Kubernetes clusters? With Kubeflow (https://www.kubeflow.org), not only it is possible but also easier to make ML workflows production ready. With recent huge demand and traction of DevOps and GitOps, many organizations are struggling to apply those practices into ML workloads. That's exact use case we want to talk about in this meeting.
Speaker:
Junaid Khan: Senior Software Engineer at Egen
Junaid has an extensive experience in software architectural designing and its implementation. His past experience covers different industries exposure ranging from public, sports, hospitality, retail and fulfillment sectors.
His previous projects involve club trainer’s activity tracking and booking mechanism, revamping from monolithic to micro-services, Distributed systems, ML optimization using Spark and user interaction applications. His current project includes working with backend micro-services based applications as well as user facing android applications for fulfillment centers.
Zoom Webinar link will be shared a day prior to the event.