Hacker News Reader

Comment by tclancy | Hacker News Reader

Does anyone have experience with the step down from this? I have zero formal CS education but two decades of experience at pretty much all types of development. I live within spitting distance of my local state university (UNH) and don’t really care about any piece of paper, but I would love to take formal/ hard CS classes, whether for credit or audit. Is there a “right” way of doing this? Does it make any sense?

View on HN

I taught my spouse to program when she decided to change fields, and did a similar thing where she wanted to master the advanced stuff by taking classes (she took them for credit on the off chance she needed it in a masters program) The classes she took were equivalent to three classes majormajor suggests:

Data Structures and Introduction to Algorithms: everyone needs this and it is maybe 50% of leet code problems

Assembly Language Programming and Machine Organization, Introduction to the Theory of Computation: this is the low level of how computers really work

Operating System Fundamentals: concurrency programming plus a lot more useful stuff

I think it is definitely worth it. Good luck!

> Data Structures and Introduction to Algorithms: everyone needs this and it is maybe 50% of leet code problems

What’s the other 50%? I’d have thought DS&A would cover everything.

A standard Introduction DSA course will generally not cover advanced DSA stuff like segment trees or Heavy-light decomposition, or theoretical stuff like number theory, or computational geometry all that stuff you would find in the competitive programming space.

Are there companies that ask a lot of leetcode bards in their interviews? That seems like overkill even for FAANGs though I can see why some groups might do so purely to filter by Math ability/interest.

From what I've seen and heard: Salesforce, Codenation, Wells Fargo, DE Shaw, so basically Hedge fund/finance or teaching/coaching/consultancy.

> segment trees or Heavy-light decomposition, or theoretical stuff like number theory, or computational geometry

I am not sure if there are any problems on leetcode that require any of these, but based on a sample of a few of their contests, if they exist they are a very small fraction (<<50%) of all problems.

The vast majority require:

- no DS&A knowledge whatsoever

- basic techniques (recursion, simple graph/tree algos, binary search, simple dynamic programming)

I do CP on various sites so I don't know if it's for Leetcode in particular. But some companies like Salesforce or DE Shaw or Codenation do include advanced problems on their tests.

I think I know more about interview questions then leet prep, but I find the remaining 50% is specific to the job but in the same vein as data structures and algorithms. So in game programming maybe it is a path finding problem or 3d math hit detection. In backend it’s an algorithm about network flow or caching. Still data structures, but more practical for the role!

The school where I earned my CS masters required those with non-CS undergrad degrees (like myself with a BS in a hard engineering field) to take 12 semester credits of additional coursework to make up for this fact. I'd say those three subjects account for about 9-10 of those credits. I suppose you could say they're important.

I did a four-year CS degree after about twenty years in industry. I learned one thing, which I do not use.

My recommendation is don't waste your time and money unless you have a specific goal in mind or want to go on to a master's program - which I also generally don't recommend unless, again, you have a very specific goal and no other way of accomplishing it.

Looking at the course listing for UNH ( https://ceps.unh.edu/computer-science/program/bs/computer-sc... ), I'd suggest auditing (I imagine that's cheaper if you're not looking for a degree?) some of the following: Data Structures and Introduction to Algorithms, Assembly Language Programming and Machine Organization, Introduction to the Theory of Computation, Operating System Fundamentals, Algorithms, Programming Language Concepts and Features, Compiler Design, Systems Programming, Formal Specifications and Verification of Software Systems. Those are the ones that look like more formal CS, and gets you into at least an intro to fun formal/theory stuff like automata, grammars, low-level programming, functional programming. Performance Evaluation of Computer Systems and Mathematical Optimization for Applications also look fun to me, or you may be interested in some of the other electives too like Computer Vision, ML, etc.

If that's local, affordable, and easy, I wouldn't worry too much about looking for something "fancier" right off the bat, I think there's a lot to sink your teeth into there. There's a lot of stuff online too on all these topics of course, but if you're like me, that's not the same as taking a live course.

I recommend you email or call the admissions team for programs you’re interested in. I did undergrad in CS and business and was interested in a different engineering discipline masters after quite a few years of unrelated work. The responses varied significantly among the programs I looked at, but the admission teams were all helpful and clear on requirements & expectations.

Why not just go through MIT's open courseware syllabus, or perhaps emailing/contacting a university teacher for guidance?

Without the rigor/ expectation of formal classes, I always wind up trailing off.

Reply on news.ycombinator.com