Follow-on to Deep Learning Specialization in late 2024/2025?

Course Q&A Deep Learning Specialization Deep Learning Resources
1

(Posting though similar questions were asked in 2022, because it is nearly 2025)

My buddy and I just finished Deep Learning Specialization. Really loved that set of courses. We are debating whether to take another course, and what to take.

Our current goof around AI project is a job application management tool for job seekers, the applicant equivalent of an “applicant tracking system” (ATS) that hiring managers use. Finding jobs that are right for you, writing cover letters, tracking hiring manager responses, etc. It helps that CS jobs are scarce in late 2024; very motivating. LOL

Here’s what caught our eyes:

  1. Generative Adversarial Networks might be useful to generate cover letters that don’t look fake, and identify when you get a response from an AI-powered ATS.
  2. Natural Language Process Specialization might be good generally, but I wonder whether Deep Learning Specialization has given us mostly what we need.
  3. TensorFlow Advanced Techniques might be good, but I worry that the Deep Learning Specialization might not have given us enough experience.

Something making our decision harder is that I can’t find any written reviews for courses on Coursera, just how many reviews there were and the average numerical rating. One option suggested in other related posts is to learn what we need by trying to implement something; we can do that, but if you are in love with a course or instructor, I would love to hear more.

(Oh yeah, and if you want to goof around with us, send me a note greening@gmail.com. The tool is in React/Remix/MongoDB.).

Any thoughts here on next learning steps for us?

2

@Greening I’ve taken NLP. While there is some overlap between DLS and NLP it is less than it might first appear on the syllabus (RNNs, etc).

One plus is it goes into much greater detail on Transformers. There isn’t really enough material in DLS to gain a full understanding, or I didn’t feel that way then. Now I (think) I finally understand them.

LLMs, though, are practically non-existent from the course. If you want, the closest you get is Karpathy’s 'Zero to Hero’s series.

That said, at the scale required almost no one is building their own foundation models today-- It is more RAG, fine-tuning, etc.

And while it would be almost useless for your applicant scan application, I started on Jeremy Howard’s (over at Fast.ai) excellent ~26 part series on building Diffusion models from scratch.

That is presently the area I am most interested in, but have had very little time these days.

Hope that helps

1 Like