If you like math, study math. But I don't think refreshing your math skills is the best return on investment if your goal is job hunting.
If you're looking for an ML job, the bar is mostly set by coding skillset and ML knowledge, which is a narrow area of math compared to what you might cover in a graduate math program. That said, it is important to be comfortable with the math that's relevant to ML.
Without direction you could spend a lot of time learning things that aren't going to help in your job search.
Logistic regression is exactly a NN with no hidden layers and a sigmoid activation function. A feedforward NN with additional layers is strictly more expressive than logistic regression.
I'm an ML engineer, and I agree with you- deep learning is by far the most common approach for new problems in informatics.
Imo deep learning is so popular because it "works". For a classification problem, if you try a linear baseline and a deep learning model, and you do a reasonable job of hyperparameter tuning and experimental design, it's likely you will outperform a simpler model. This holds true across many problem spaces.
I think the issue is that modern DL frameworks make it a little too easy to get pretty good performance on new problems. Other techniques generally require more background knowledge to make reasonable modeling assumptions, and still frequently perform worse than a naively applied DL approach.
I think DL will remain, in practice and education, a very popular tool. But it is essential to learn traditional statistical inference and other background to appropriately contextualize DL models so it isn't just some form of black magic.
I have a couple Google Home mini's in my apartment, an android TV, and almost all of my lights are Phillips Hue. I use the assistant constantly to turn on/off lights, change colors, dim them, play things on the TV, etc. Very high usage.
I think using the assistant for things like calendar, weather, traffic, calls, etc is more difficult than just using a phone.
But being able to turn off all the lights in my apartment when I'm ready to go to bed by just saying "lights off" is really nice.
At Joe & the Juice in Palo Alto (cashless), I was buying a drink with my friend, and a clearly homeless man in front of us was trying to buy a coffee with cash. They refused and he got upset, so my friend paid for his coffee to de-escalate.
We were outside when the cashier ran up to us in the street, explaining "Please don't buy drinks for these guys, they come in here and get a drink and use that to stay all day. It's not the look we're going for, and it's one of the reasons why we have this cashless policy."
Cashless stores are convenient and I understand the arguments for them (as well as the arguments for not buying homeless people free stuff), but as long as cashless can be used to disenfranchise lower income people, I'm opposed to it.
I don't think this is surprising, but I'm also not surprised that VCs and accelerators still trend toward younger founders.
It makes sense to me that older (successful) founders likely leverage their own experience and network to a greater extent, but this isn't necessarily what seed stage investors are looking for. I think investors are more in the game of betting on people/teams, injecting their own experience and network as leverage. Older founders have less need, therefore less leverage- and older founders without these support networks are poorly positioned anyway.
Theres a big difference between running a successful small business and architecting a tech startup.
If you're looking for an ML job, the bar is mostly set by coding skillset and ML knowledge, which is a narrow area of math compared to what you might cover in a graduate math program. That said, it is important to be comfortable with the math that's relevant to ML.
Without direction you could spend a lot of time learning things that aren't going to help in your job search.