I was contacted by a recruiter after visiting Google at a career fair. I then went through two phone interviews with live coding focused on algorithms, data structures, and optimization.

After that, I was invited on-site for a full-day interview (Google College Day), where I had three interviews.

The two phone screen interviews were pretty basic. It was a while ago, so I don't remember everything. The first few questions were about the Fibonacci sequence with optimization, the lowest common ancestor with optimization, and finding a repeated number with optimization. I suggest you start out with a naive answer so that you can at least solve the problem, then think of a better way to do it.

The on-site interviews were really hard and intimidating.

The first interviewer asked me about OO concepts with some trick questions.

The second interviewer was totally unprepared and just kind of thought of questions on the spot. He asked me about designing a server and how to prevent failures, how to reroute requests, and making backups, among other things.

The third interviewer was nice but tough. He asked me about Bloom Filters and techniques used in URL whitelisting in Google Chrome. I had no idea as an undergraduate.

Find a repeated number in a list of numbers.

Write a function to generate the Fibonacci sequence.

I have a superclass A and a subclass B of superclass A. B overrides a function foo() in A. I have an instance of class B. How do I call foo() using class A's version?

How do you check if a URL is bad really fast on a Google server? The point is for the user not to notice the lag in the checking.