Ask HN: Which industry roles actually require daily use of CS knowledge?

3 points by mcrwfrd 4 years ago | 11 comments

I recently started refreshing my knowledge of data structures and algorithms and discovered that I actually really enjoy the subject. Studying the Union-Find data structure, for example, has really helped me rediscover the joy of programming.

After a few years in the industry, I have mostly accepted that most programming jobs these days are "lego programming", in the sense that we don't really use "computer science" knowledge in our day jobs, we mostly just connect API's and write business logic. I've mostly come to terms with this. But a couple of times a year, I start to wonder if there are any jobs these days that are more than just "lego programming".

So the question: does anyone here have a job in which the code they write is more "computer science" and less "lego programming"?

My hope with this question is to have a few modern examples of interesting computer science roles in industry, rather than reignite the old "all we do is CRUD" pity party.

PaulHoule 4 years ago |

For me the phrase "Computer Science" is considered harmful. They call us "software engineers", but people see "science" as having a higher social status than "engineering".

Part of "computer science" is about particular accidents in how we build computers, such as CMOS logic or the 8-bit byte or the bizzare ways that C overloads keywords, but there is another part of it which is universal. There is intelligent life all over the universe that lives under the ice on moons of gas giant planets and if they have built computers they know the results of Alan Turing and Kurt Gödel under some other name. They know that you can't sort a list with fewer than O(N ln N) comparisons.

It's important not to get so full of yourself that you look down on the work you are doing. I see the word "science" in data science being corrupting because many "data scientists" think they are too good to practice software engineering, to put effort into repeatable builds. They could put their work on wheels if they built a script that could reliably build a machine learning model or generate the monthly sales report every month but many of them seem to think they are too good to do that and it diminishes the value they give to the organization.

Organizations too don't see the value that "computer science" could give to an organization. I worked for a web consulting shop that was in crisis particularly because it had invested in a proprietary programming language that was no longer being supported. They probably could have hired a graduate fresh out of school who'd just taken a compiler class and built a transpiler that would compile it to P.H.P. or some open language, but management didn't know that, didn't believe that, and now it is too late.

the_only_law 4 years ago | |

> but people see "science" as having a higher social status than "engineering".

I wonder why, I mean generally engineers are paid more than scientists, specially towards the bottom of the respective fields and money seems to be a big driver of social status.

dusted 4 years ago | | |

The reason engineers are paid more is that they're a safer bet for moderate(to high) value creation, while actual science is not guaranteed ROI in any reasonable time frame.

The reason science has a higher social status is probably because most passionate intellectuals agree that it is indeed nobler and more important work to discover fundamentals and increase the sum of human knowledge than to increase profit margins of multinationals.

theamk 4 years ago |

There are are lots of them. OS/database design, new computer graphics, robotics, embedded. For example, I just checked front page of HN -- there are 2 submissions which involve non-trivial CS algorithms: "you.com" search engine, and "ancient secrets of computer vision" course.

Even if the primary output of the organization of the website, there are still places for algorithms -- for example one of my friends is involved with bringing data science algorithms into "production" -- while initial experiments can be done in the Python with Jupyter notebook, you often need to rewrite the algorithms into high-performance language of your choice.

ThrowawayR2 4 years ago |

Yes, most programming is Lego programming but the parts of the industry that require deep CS knowledge are the ones that build the Legos. The jobs are out there, though they are harder to find: OS development, compiler/ language runtime development, hardware related jobs like low-level embedded systems and driver development, games development, database engine design, FAANG-level distributed systems, etc.

the_only_law 4 years ago | |

They also seem to tilt much heavier towards the > senior level roles, with many I’ve seen demanding a decade of domain experience.

I imagine that most lower leveled positions are sources from university pipelines.

aprdm 4 years ago |

You want to look for jobs outside the fullstack/web space . Look for engineering companies that aren’t tech companies.

mettamage 4 years ago |

My 2 cents:

High frequency trading. The hours aren’t great.

the_only_law 4 years ago | |

How does someone which isn’t a university student in the internship/university hire pipeline in a financial hub and has no prior experience in the domain get there, assuming their skills up up to it.

mettamage 4 years ago | | |

Hmm, my best bet would be: make a small trading project, do something low latency. If you're CV is ragged, then make that your tagline (including Github link) and explain the situation.

yuppie_scum 4 years ago |

The SRE/DevOps space makes a lot of use of structured data (yaml, json) etc.