Interview preparation
The assessment regime that shapes most of what this spoke holds. Two of three founding sources exist to get someone through a technical hiring process, and say so.
The claim: syllabus follows assessment
coding-interview-university is a computer science study plan that explicitly excludes frontend, full-stack and SQL. There is no account of computer science on which those are peripheral and NP-completeness is central. There is an obvious account of the interview on which exactly that is true.
awesome-system-design-resources makes the framing explicit rather than structural: a section titled “How to Answer a System Design Interview Problem,” a problem set graded easy/medium/hard, and onward links to courses sold on interview readiness.
So what a very large number of engineers study is determined by what a hiring process tests — and the hiring process is not represented in this corpus at all. Every source here is written by a candidate or for one.
The assessment, distilled (2026-08-05). llm-interview-questions-hoang is the cleanest instance: not a curriculum shaped by an exam but the exam itself, 50 LLM questions with model answers and no learning path around them. Where the two founding sources let you infer the assessment from the syllabus, this one prints it. And it shows what the LLM interview presently rewards — recall of mechanism: define tokenization, state the LoRA-vs-QLoRA difference, explain attention. The flashcard, not the capstone. It is also the spoke’s first look at the AI/LLM-engineering interview, where coding-interview-university and awesome-system-design-resources cover CS fundamentals and system design.
What that predicts, and what would falsify it
If the thesis holds, a discipline with no standardized interview should produce differently shaped curricula. ai-engineering-from-scratch is the one such case the spoke holds, and it is built around construction and capstones rather than recall and whiteboard performance (synthesis).
One case is not evidence. The falsifier is cheap to state: a construction-based curriculum aimed squarely at an exam, or a curated-recall curriculum with no exam behind it. Either breaks it.
What is missing
Nothing here says whether these interviews predict job performance. That is a real research literature and the corpus holds none of it — recorded as growth edge 3 rather than assumed either way. It matters, because if the assessment is poorly correlated with the work, then the curricula above are faithfully teaching several hundred thousand people to the wrong target.
Related
meta-curriculum · coding-interview-university · awesome-system-design-resources · ai-engineering-from-scratch · llm-interview-questions-hoang