Coming from an experienced industry leader in Applied ML and ML Engineering, the session will look at various definitions and contextual semantics for MLOps, look into the different MLOps components, look at the critical area of model serving in detail and also discuss several case studies.
The presentation will highlight the various reasons for the chasm to production for ML models, articulate the importance of Dev-Prod reproducibility, look at various industry definitions of MLOps and present maturity of MLOps across different generations. The MLOps value-map and how different industry players are filling in the value will be shown. This will be followed by explaining important concepts of MLOps for Applied ML practitioners such as different types of data drift, model drift and concept drift. Model Serving will be described in detail from both batch and real-time angles, including concepts of scaling and real-time API.
There will be a section on careers in MLOps as well as different case studies in the industry highlighting the above concepts
Key Takeaways:
- Understanding MLOps and its components
- Theoretical Concepts in MLOps explained
- Model Serving – in both batch and real-time contexts
- Careers in MLOps
- Real-world case – studies
