Atomic Talk Title: Shift-Left Testing in Machine Learning: Enhancing Model Quality Early

This topic underscores the crucial role of Quality Assurance (QA) in the Model Engineering phase of machine learning development. By involving QA early, teams can conduct over-fitting and under-fitting tests, drift testing, analyze cluster scores, and fine-tune model parameters. Integrating these evaluations at the outset helps manage functional issues and elevates the overall quality of machine learning solutions before entering the actual validation phase.

Talk Takeaways

• Recognize the importance of shift-left testing in the ML development process.
• Understand QA’s contributions during the Model Engineering phase.
• Learn techniques for early detection of over-fitting, under-fitting
• Discover methods for analysing cluster scores to tune model parameters.
• Appreciate the benefits of early issue detection on the quality of ML solutions.