Pros
Put simply, NSF is an agency that receives grant proposals and brings in panels of experts to evaluate the merit of those proposals for funding. There are 7 scientific directorates, a few offices of the director, and offices that deal with HR, Budget etc. Note that no science is actually done there (other than data science). There are frequent, good lectures by grantees that are visiting the agency. You get exposure to the world of scientific research and what's being done, science funding, and to a large extent, academia.
Cons
Staff at NSF are highly stratified, and the culture is something of a socio-educational caste system. There is an upper class - the scientists, and a lower class - the "administrative" or "support" class. Few in the agency occupy the space between these two poles, with some exceptions for budget and accounting staff. All program directors (PDs) hired at the Foundation are PhDs, and most are tenured research professors. In the past there were some non-PhD PDs, but this is no more. Even assistant PDs, of which there are very few positions, are now PhD requiring positions. Because PDs are evaluating and discussing scientific research, it makes sense that they are PhD researchers themselves. What I found troubling though, was a pervasive sense that not only were PhDs the only ones that could understand science and research, but the only ones capable of making any decisions of any import or allowed to take ownership of anything meaningful. Administrative staff for the most part deal with the highly repetitive tasks related to processing the massive numbers of proposals NSF receives each year, and a ton of travel. There is travel to bring to NSF the numerous panelists who discuss and evaluate grant proposals. Additionally, NSF staff scientists travel frequently to conferences, and there are many visiting scientists who work at NSF for 1-2 years and travel back and forth frequently to their home institutions. Travelers pay their own costs and get reimbursed by NSF. Processing travel is a punishing task on several levels, from people's frequent fussiness about their travel plans, to applying for financial reimbursements for travelers and being scrutinized by the budget office as though you are trying to embezzle from the agency. When reimbursing panelists, we had to get their email confirmation that the amount was correct. If you later discovered that a rounding error would cause the reimbursement to be one cent more, you had to ask their approval again. Quite humiliating. Another major task for administrative staff is ordering refreshments for the panelists, tidying them up throughout the panel, etc. The highest tiered administrative staff get to do work of a more analytical nature - querying databases built from grant proposals and awards. Much of this analysis is based on the demographics of who is applying and who is getting funded, which can become repetitive, but there is some challenging work in this area that at least requires some cognitive activity. It is worth noting that though there are many female scientists at NSF, the majority are men, and especially in the traditionally male dominated fields of science. Most of the administrative staff, the servant class, are women. Scientists at the agency are constantly talking about how to increase participation in the sciences by women, minorities and underrepresented groups generally. I found this profoundly ironic, because NSF seemed to me an almost perfect model of the kind of organizational system that keeps people out/down based on their pedigree. This black or white thinking paradoxically permeated the culture created by the learned leaders there. Most of the scientists at NSF are fantastic; they are teachers who are awake and alive, but the ones that are jerks can be incredibly condescending, inconsiderate, and entitled. For them, micro-aggressions (and macro) are behavioral norms. Something about NSF’s caste system also seems to increase unhealthy competition among the administrative staff. I think this is especially true for the males who find themselves in the lower, female caste. I found several of the male analysts particularly aggressive and overbearing, sometimes combative. So bottom line – if you are a science-interested person attracted to the science part of NSF, but not planning to go into research, I encourage you to go to a place that applies science. If you are not particularly interested in science but want to be seen and treated as an equal, I encourage you to go elsewhere too. For me, the constant reminders of being of the lower caste (like being asked if I was the helper, hearing a scientist describe the scientists as the “A” team and the administrative staff as the “B” team, etc.) were challenging to my sense of worth as an employee. I saw this in others too, and I think it can discourage people from trying to reach farther. I have gone to another agency and it has been wonderful. The staff come together as a team to do great things and sure, we are on different levels, but it’s a pace where I can grow based on my own initiative and skill. I think we all deserve that. I wish I would have started in my new agency sooner, because in addition to feeling better, my experience would mean something. Agencies are complex and you often need time to understand the whole and how all of the pieces fit together to perform optimally. If you go to NSF, learn and enjoy as much as possible, it may help you get into the federal workplace, but have a plan for getting out, and don’t wait too long.