Why not go back to HFT or more general quant trading? In my experience the direct link between work done and profits or lack thereof keeps the skill levels high and reduces internal politics.
microcorruption is great. I originally signed up in 2015 and completely bounced off the first level after the tutorial. I worked through the levels on and off after that, most progress made during a ski trip where I injured my knee badly enough to prevent actually skiing.
Judging by notes I made it as far as the penultimate level of the original set in 2018 and got stuck there, it's a shame the new site doesn't show your submission dates, which I think the old one did.
Although my progress was ridiculously slow, am quite proud that I just sat with the problems when stuck without looking up the solution. Maybe this year I will actually finish it.
Cython is pretty excellent for this sort of thing. The article contains a link to a Cython solution written by Travis Oliphant showing how concise (and fast) it is.
I just used Cython to speed up a non-vectorisable bit of code in a Gibbs sampling algorithm and I was impressed by the performance increase and ease of use.