I’m in the process of transitioning from my current career in teaching to the NLP career via the Python path and while I’ve been learning on my own for about three months now I’ve found it a bit too slow and wanted to see if there’s a good course (described in the title) that’s really worth the money and time investment and would make things easier for someone like me?

One important requirement is that (for this purpose) I’ve no interest in exclusively self-study courses where you are supposed to watch videos or read text on your own without ever meeting anyone in real-time.

  • simplymath@lemmy.world
    3·
    3 days ago

    I work in the field. Generally, jobs that include AI development generally require advanced degrees and the vast majority require a PhD with peer reviewed publications in major conferences. You will be fighting an uphill battle if you don’t have an advanced degree in mathematics or computer science. You also need to know calculus, linear algebra and statistics to understand how modern machine learning models work.

    In short, while online courses can be perfectly effective, unless they’re through an accredited higher education institution, I don’t think it will help you compete with other applicants who have 8+ years of schooling and published papers.

    That being said, Georgia Tech and the City University of New York both offer master’s degrees in data science via remote master’s programs where the courses happen after work hours and are meant to be completed while working full-time.

    • SurpriZe@lemm.eeOP
      1·
      2 days ago

      Could one take these Master’s online courses if their Bachelor’s major was in a completely unrelated field? Well, I did hear that my Linguistics degree could he used for NLP but it’s still not a direct path.

      • simplymath@lemmy.world
        2·
        2 days ago

        Yeah. the CUNY one is definitely meant for career switching, but the Georgia tech one will probably expect you to know the math fields listed above as it is fairly competitive. Though, I know someone with an Economics bachelors who did quite well in the CUNY program. They even offer an introductory course for people with 0 programming experience. I really do think it would a good fit, given your background. Link here. A head’s up though-- graduate degrees will require more independent work than undergraduate did. Like, course meetings were less lectures explaining new content and more answering specific questions after you learn the content on your own. I was expected to have completed the homework before the topic was covered in class (though it wasn’t graded for correctness). I would say that’s the categorical difference with advanced degrees.

        I’d spend some time on Khan Academy to brush up/catch up on the basic math concepts. That’s where I learned those topics.

        I know you asked for some kind of personal interaction, but that content is the gold standard for math education. You can always ping me if you have specific questions and I’ll do my best to respond.

        • SurpriZe@lemm.eeOP
          1·
          6 hours ago

          Thanks a lot! I appreciate the effort you’ve put into the responses. In general, what concepts would you expect an entry level NLP job to require to know from its applicants? I think I’ll stick to learning on my own with an occasional tutor/coach down the line.

          • simplymath@lemmy.world
            1·
            2 hours ago

            I’m not sure there are “entry level” NLP jobs that aren’t for sketchy startups without a future. I can’t imagine using openai APIs is a stable career as much as knowing how to make them.

            If you were a senior programmer, there are roles that don’t need so much math – DevOps, reliability engineering, cloud deployment, and systems administration are some keywords you can look up on LinkedIn. These are quite broad and would be available at basically any company with a website or app.

            Personally, I got an undegrad degree in applied mathematics and only kept going to school because the roles I wanted (ML development) required it. Your background in linguistics could certainly be useful for a team in which other people are the coding and math experts, but I do not come across those roles on LinkedIn very often outside of academia, which, unfortunately, require the degree.

            Its really good to hear that the Andrew Ng class is comprehensible to you, so I think you’re on the right track either way.

            In lieu of a degree, the only other thing I can suggest is building a portfolio online (GitHub, probably) and maybe contributing to an open source project. It will be very hard to find roles that don’t require you to know git and contribute to complex software projects, whether or not you have a degree.

  • nickwitha_k (he/him)@lemmy.sdf.org
    1·
    3 days ago

    I recommend the edX CS50x course. It’s a bit more general about computer science than your ask but, it’s free, an actual Harvard/MIT course, and covers algos, data structures, and programming in C, Python, and JS (last I checked).

    • SurpriZe@lemm.eeOP
      1·
      2 days ago

      Thanks. Is there any proof like a unique code you get at the end of the course proving that you’ve completed it?

      • simplymath@lemmy.world
        3·
        2 days ago

        Yeah. They do, but the certificates aren’t free and wouldn’t necessarily contribute to a degree anywhere, so I’m not sure what the advantage would be of doing it that way.

        To give you a sense of the level of content, here’s an introductory class in ML.