Couple of great and interesting points. I'm going to relist each one to address.
1. Building and maintaining a nuclear powerplant alone is not a job you can do with 500 people.
That's not entirely true. Building one, for sure, that would take considerable resources, people, skill sets, etc. But running one, that takes very few people - especially for the low output nuclear reactor a society so small would need (as you so very accurately point out).
2. [W]hy would you even run something within the ballpark of 1GW output for less than 100k people?
Good question and excellent point. You wouldn't need a lot of power right away, but you might need it in 100 years. There are some very simple designs that could last 100 years (given proper maintenance, which I'll admit immediately might not be available in this scenario). And if a full small-scale nuclear reactor won't work, there are other varieties of nuclear-ish power (@Skroe would kill me for that reference) that might work even more effectively, for the requirements. The
Radioisotope Thermoelectric Generator is a good example. Tons of output for hardly any maintenance.
But why indeed. I think the reason is because if you can have more output than you need, rather than need more than you have, and you can maintain it given the scenario, it's worth having.
3. [B]ut from there it's all about getting some reliable stuff you can actually produce and maintain, most likely going back to coal.
If I understand the basics of coal, it's a very labor inducing process. I don't think that would be an effective source of energy given how few people would be available and how many more tasks would need attending/developing.