I don’t know how this would work for CS, but in law school I had one professor that did final oral examinations as follows: there were about 15-20 cornerstone t

I don’t know how this would work for CS, but in law school I had one professor that did final oral examinations as follows: there were about 15-20 cornerstone topics from the class that were possible exam questions and were announced a few weeks before finals. Then, during finals week(s), each student had a 30 minute block during which they would randomly select 3 of the 10 topics by pulling note cards and engaging in a conversation about the topic with the professor. If you knew the basics of the topic you were almost guaranteed a B or B+; if you demonstrated novel or expansive knowledge you got a higher grade… and you really had to not study to get a C+ or lower. There are flaws in every method of measuring competency but I thought this was one of the better ones, and it is very difficult to cheat on. His exams were by far the least grumbled about and I don’t recall a single student that thought the method was unfair. And even though this required more upfront work from the professor he said it was ultimately less time consuming than grading written exams, and far more enjoyable.