Software Engineering Institute

Our one-of-a-kind tool and process, Machine Learning Test and Evaluation (MLTE, pronounced “melt”), can be used to evaluate machine learning (ML) models from inception to deployment.

Designed for interdisciplinary cross-team coordination, the MLTE framework defines a testing process and offers a supporting tool to allow teams to more effectively negotiate model requirements, evaluate model functionality, and share test results with all stakeholders. Providing various artifacts to support the processes of developing as well as testing and evaluating ML models, the MLTE tool includes negotiation cards, quality attribute scenarios, a test catalog, and a reporting function.

Overall, the MLTE process and tool increase ML model production readiness.