Accelerating the deployment of predictable ML in production