EBoard 11: More interview practice

Approximate overview

• Administrivia
• Activity
• Debrief

Administrivia

• Next week we’ll have a founder of AppDev visiting (remotely).

Questions

How do I find out about REUs and more?

CSstudents mailing list.

Sam has links on his summer opportunities page

Activity

• Five minutes of reading and trying to understand the question you will ask your partner. (There’s also some prep you may want to do.)
• Twenty minutes of Red Interviews Black.
• Five minutes discussion with each other. “Here’s what you did well.”
• Twenty minutes of BLack Interviews Red.
• Five minutes discussion with each other. “Here’s what you did well.”
• Back to the main group to debrief.

Debrief

For the GrinCo problem (also the Waymo problem)

• Intentionally open-ended. (Perhaps a little too open-ended; I should have specified number of cars and the time cost of getting from place to place; or perhaps not, since there are bigger issues at play.)
• What are the goals I’m trying to meet?
  • Minimizing average wait time.
  • Minimizing energy use.
  • Maximize “in use time”
  • Maximizing profit.
  • Avoid certain locations.
• “Do you have historical data and how consistent are they?”
  • If you do have historical data and its consistent, your answer might be “ML”.
    • It would be nice to know what the policies ML comes up with are.
  • If they aren’t consistent, you need to be a bit more creative in designing policy.
• Some ideas worth sharing.
  • How do you decide who to get? (Assuming only one car available.)
    • Nearest
    • Shortest trip of “here to pickup to destination”
    • The one waiting longest
    • A formula based on all of those.
  • How do you decide which car to send? (Assuming only one request.)
    • Closest?
  • Where do cars go when they aren’t engaged?
    • Designated garages.
    • Distributed
    • To where we expect people.

Elevator problem

• What is the UI for people waiting for the elevator?
  • Up/Down from where I am.
  • I want to go to floor X (only available in the lobby)
  • ???

For the MegagoogazonbookInc problem

Suppose you build a hash table: How do you efficiently figure out which color each key in the hash table is closest to?

• Think of it as a graph problem.