Machine Learning Technology

Question

I'm currently in college and about to start university in 2024. For my course weere currently programming in C#. I'm most competent with C and am proficient with Python. What programming languages besides are often used in industry for machine learning applications or projects? Some people have recomended C++ or even Rust I'm more so leaning towards C++ due to it being more widely adopted. Should I be focusing on C/C++ and python; or would learning rust be useful?  Besides programming languages what relevant technologies are used in industry as well.

crescent · Accepted Answer

If you're thinking about career prospects, most companies won't care too much about which languages a junior has experience with. Not the worthwhile ones, anyway. Experiment, see what you like and don't like, you can't do it wrong.

As for more tooling; version control and automated testing (unit testing) will add a lot of professionalism to how you work. Version control will definitely make learning languages easier when you know you can undo mistakes. I know these aren't ML specific but all developers should use them.

More importantly, look for summer internships and consider a placement year at university. It just makes job searching that much easier once you graduate. Plus, you were going to write code all summer anyway, right? May as well be paid too!

Good luck with your exams!

mmommommommo · Answer

I would say mostly python and c/c++ but it would also be beneficial for you to be familiar with these languages ecosystems. In the python world it’s good to have hands on experience with libriaries like numpy, pandas, pytorch, tensorflow, scikitlearn and matplotlib (there is more but it’s good start for the young ML engineer ;) )

Sebastian Cattes · Answer

There are two parts of Machine Learning
1. Python/R/Julia/some other higher level interface
2. C/C++/Fortran/Haskell/some other low level high performance computing language

I would say 80% of ML is done in Python/R/Julia where people have built smart packages on top of the lower level languages to do machine learning. Numpy, Tensorflow, pytorch etc. all have a huge portion of their actual code in C++ or sth like that and the python is on top to allow you to actually use that code properly. Most people never touch the C++, you don't need to.

There are companies for whom the overhead and performance loss of python is high enough that they directly use the c++ XGBoost implementation instead of the python/r package. High Frequency Trading for example.

So I would recommend to stay with python and C++ (maybe with and 80/20 split) and mostly use python. Learn some statistics and how to use the packages @mmommommommo recommended. If you wonder how something works or how to add functionality for a ML package, you will then end up in some C++ code where you have to add the new functions to bind them into the python package.

Have fun learning :)

Billybones · Answer

Forget C++. Starting to learn it now is like starting to learn how to punch holes in Hollerith cards when higher level languages already emerged. With ChatGPT exponential growth the next chapter is going to be prompt engineering. Look for those opportunities

WaigongCain · Answer

Freshies, always gotta watch out for D1 kiddo. But good luck with schooling, make sure to ask bunch of questions and do a lot of networking.

Machine Learning Technology

About

Software Engineer

Members