I am sharing my experience interviewing for a Software Engineer position at Meta. The interview process consisted of a phone screen.
The first 15 minutes included standard behavioral questions:
The technical portion involved implementing a median image filter function. This function takes an input and output image, kernel width, height, and width as parameters. It must write every output pixel, where each output pixel is the median of a k-by-k square of input image values centered at the corresponding location. I walked through an example, confirmed my understanding, and broke down the problem into a two-step approach: a helper function to get the kxk grid given an index and a function to compute the median. I assumed sorting was done for the grid and provided the runtime and space complexity. My code had some bugs, but the interviewer was satisfied.
I am now moving on to the onsite interview loop.