Tell me about a time you had to deploy an ML model under a tight deadline at Grainger. How did you balance speed versus accuracy?
