Yandex

One-hour video interview, where I had to solve a medium-level algo problem. I'd solved it, yet there were bugs in my code, so I spent another half of the interview trying to find them. Finally, I'd found them all, but time came to the end.

Improving existing algorithms and creating new ones involves several key elements: * Machine learning * Working with very large amounts of data * Writing algorithms to work with them An example of this is image search optimization in Yandex.i

It used to be a four-part interview: * Screening * Java/Scala section * Databases * Algorithms Now, it's much simpler, with just two sections: * Theory * Practice (solve some problems in pseudocode)