Why is MLOps Hard in an Enterprise? - Challenges and Solutions

Explore the challenges of implementing MLOps in enterprise organizations through this insightful podcast episode featuring Maria Vechtomova and Basak Eskili, ML engineers at Ahold Delhaize. Gain valuable insights into the complexities of data ecosystems, cross-team collaboration, and the need for standardization in ML tooling and infrastructure. Learn about Ahold Delhaize's approach to model reusability across brands, the evolution of their MLOps practices, and strategies for overcoming common obstacles. Discover the importance of bridging the gap between data scientists, infrastructure teams, and IT departments, as well as the critical aspects of model operationalization, monitoring, and cost management. Delve into discussions on guardrails, credential sharing, the differences between MLOps and DevOps, and the potential of AI and ChatGPT in new projects.

[] Maria & Basak's preferred coffee
[] LLM's in Production Conference Part 2 coming up on June 15-16!
[] Maria & Basak's background
[] Takeaways
[] A colorful history
[] 4 levels of evolution
[] Standardization and Model Registry Evolution
[] Ahold Delhaize Standard task
[] Ahold Delhaize Workflow
[] Avoiding tooling sprawl
[] Guardrails
[] Secret sharing and credential sharing sloppy processes
[] Distrust between devops engineers and data scientists
[] MLOps vs DevOps
[] Monitoring pieces heroes
[] Future accumulative cost issues
[] Exploratory phase in notebooks
[] Production level code system
[] Blind spots
[] Using AI and chat GPT on new projects
[] Wrap up