It is super helpful to have a working knowledge of common data structures- most algorithms are optimized for data in a particular format.

So, probably the single most important one for me has been the concept of immutable data structures: whether in C, C++, Java, or Javascript, treating my data immutable has made debugging things a lot easier, especially in environments with parallel execution and shared memory.

The next one would be abstraction of storage. Whether using an array, linked-list, or something more exotic, I usually want to hide that from the rest of my program if possible. In general, abstraction is a very handy tool--though you are going to screw up and overdo it until you've got some experience (mainly gained by screwing up).

Next top one would be recursive functions to operate on tree-based structures--though I hesitate to use them automatically in languages that don't have mandatory TCO unless I know I won't blow the stack. So many things are plain elegant when you write the problem recursively.

Those are three general principles of engineering and design that have served me well. Some specific things that are useful are: the general idea behind finite-state machines and when to recognize if they apply to your problem, regular expressions and their use, pitfalls of the floating-point representation of numbers, basic linear algebra for graphics and simulation.