What does it take to not be able to get an engineering job as a computer science major in Silicon Valley?

The places you want to work and I pass on candidates with computer science degrees that lack the basic aptitudes (we can't seem to teach them) they should have needed to earn computer science degrees but don't in a lot of places because that leads to a high drop-out rate (Joel Spolsky observed a 40-70% rate in the 1990s, and hiring in industry for one company subjectively we went from a negligible reject rate to about 50% after a change to a more lenient data structures professor at one school) with lower revenue for the computer science department. I've worked places where senior management enforced a lower hiring bar and found that candidates lacking the aptitudes generated more work for other people than they put into the product making them net negatives so I won't budge, although I do compromise on experience because the right guidance can produce good results using junior people with aptitude. Those aptitudes are 1.  The ability to form consistent mental models Saeed Dehnadi has a paper on this one http://www.eis.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf with the title referencing the pair of bumps in place of a bell curve   seen in an introductory programming class and mentions recursion and   concurrency as other aptitudes. 2.  Thinking through problems logically, identifying their end/edge conditions, and expressing a solution in code. An   example of this might be the classic "reverse a string in place" job   interview question.  People fail to get to start at the ends and swap   until getting to the middle, not stop in the middle so they have the original string, or don't handle empty strings.  Extreme examples with computer science degrees can't code a solution to that scope of problem given an English algorithm description and illustrations. 3.  Applying knowledge to engineering problems Some people learn by rote.  They can parrot back numbers and information related to a context but not apply it to problems.   For example people can rattle off big O complexity of all their data structures but not choose one for simple problems. 4.  Indirection People can't think of when to use it, like from part of one data structure to another.  They get confused between pointers and what they refer to in simple linked list problems. 5.  Concurrency Even given definitions of lock, unlock, and semaphores or monitors some people can't think along two lines of execution and come up with the right ordering to solve a simple thread-safe producer/consumer problem. 6.  Recursion

Here is my perspective as a computer science professor (and a software developer in Silicon Valley long, long ago): A number of CS majors are in it simply because of the (putative) job prospects.  They have no passion for computers, and thus no intellectual curiosity.  They go through their coursework obediently, but never give any thought to questions like "Why are we doing it this way?" or "What could else could I do with this method?"  They have no idea that a lot of the real learning about CS comes on one's own, not through formal coursework or if related to coursework, not specifically required by the professor. These students worry only about "Will this be on the final exam?"  I don't mean to say that that is an unimportant (though I don't give finals), but if that is ALL a student thinks about, the consequences are obvious.  My favorite example of that is use of debugging tools.  At my school, these are taught as part of the beginning programming course, but I've noticed that most students never use debugging tools after that; it was just something to learn for the final.  It never dawns on them that the debugging tools are there to help THEM, to save THEM time and frustration.  Unfortunately, at many universities (or better, with many specific professors), "obedient" students like this can do quite well in terms of grades.  Some professors give "practice tests," for instance, which turn out to be close to the real thing.  Other professors will give detailed outlines on how to do a programming assignment.  It is quite possible for a CS graduate to have never written a complete program from scratch in his/her entire college career.  At research universities, many professors minimize the time they put into teaching, and thus recycle old exams, old assignments and so on, again making it easy for the uninspired to get good grades.  Often a large portion of the grade will come from projects, which students can do well on by asking lots of questions of their professor, TAs and classmates. If you want a rough but quick test to see if a new CS grad has any insight, you don't need to ask him/her to write recursive code (though I agree it's a good test).  Instead, just ask, "Have you ever written a program for your own use, not for a class?"  If they say no, they are almost certainly in the "obedient" group. Note, though, that any kind of test becomes known and circulated, and thus gamed.