The first step was a 30-minute web test, which consisted of graduate-level software engineering questions.

After passing that hurdle, there was a one-hour phone interview based on a code fragment that was quite difficult. The interviewer was not friendly, and the process did not seem to go well.

The code fragment was C++ code (not quite compilable, but almost so). The code looked simple, but the questions delved into how one might create data structures to compile multiple-inheritance classes and, in particular, what the class memory layout might be. This seemed to be a test to see if one could design a compiler on the fly.

Thank goodness compilers aren't actually designed that way.

Yet another company thinking they need folks right out of graduate school, in my humble opinion. I hope that works out for them.

Given two base classes and a class derived from them, how might one lay out class instance memory so that polymorphism will work correctly? The problem was simplified in that one could assume memory was a series of same-sized slots, and methods and attributes took up one slot each.