There are tons of mistakes in these models, many are more subtle. Which is why people should stop hyping it so much. It's not just math..

That was the Skylon idea, but it's not efficient and very complex.

Rockets do not need "air", they need oxygen.

I agree.

All subs related to AI or ML appear to get flooded with this stuff right now.

Why should they? They haven't presented a working prototype of their engines in 30 years, and there are a lot of questions about that design.

And even what they propose to do with it (Skylon) isn't really a good idea.

>with newer technologies towards single stage planes

Single stage to orbit from earth is a waste, whatever great technology you use. If two stage full reuse works out it will always be more efficient.

One problem of many open-source models is that they are badly documented (there was a big discussion last year that many models used in scientific papers couldn't be replicated).

So reverse-engineering them is often harder than building your own from scratch.

>from point of view of getting a job

How can anyone answer that? "a job" as in "any job"?

Yeah, python is probably ok, don't know, depends. I will depend on the job requirements.

You are not going to get an "ML job" just because you know some python. I mean you may be hired as a python developer as part of a project using ML, but you need to know much more than that to do real ML work.

What you "need" to know will depend what project or job you are working on. Unless you are working alone or start from scratch it will probably not be your choice anyway.

It won't hurt to know more than one language.