Interview question asked to Technical Program Managers interviewing at Hulu, VMware, Uber and other companies. Original question asked: What is the mathematical and architectural approach behind generating highly reliable arrival time estimates for millions of concurrent drivers?.