This is a good one. I was actually just asked this by one of the Lyft interviewers on preptick.com where I'd reached out to her for practice before my real interview. Here's how we went about the answer - Clarify the question: What's the scope? Coming up with an ETA estimate of drivers for rider requests? Coming up with an ETA tooling which lets riders inform drivers when they'll reach? This is important, it completely changes the nature of the question. Let's assume it's the former. ETA estimates. My MVP solution would be to use Waze or Google Maps' API and let that do the heavy lifting of the calculation. And surface this information in real time to the rider, to see how it adds value, and whether riders find it useful, and whether the relevant metrics show us this. If the interviewer precludes using 3rd party tech (my preptick Spotify interviewer insisted I come up with my own solution) - I proposed building a shortest path algorithm to figure out the quickest distance, where I'd use graph structure, a depth first search, to diagnose the quickest route. The length of each branch would be the time taken to get there. For a rudimentary, barely-viable solution I'd estimate cars within the city to travel at 30km/h, and on highways at 90km/h on average to estimate how long this time would be. Having said that - this answer will completely change based on where the interviewer leads you, so I recommend doing a mock on someplace like preptick.com so you can get a realworld feel for this kinda stuff.