An important caveat to the vaccine-latency estimates is that some people will have been infected before they were vaccinated, but still be in the incubation period. These people will not be protected much by the vaccine, obviously, but in the graph they look like they were infected post-vaccination. So when estimating the risk others pose to you, subtract off the approximate incubation period (~7d) from the latency estimates. (But when estimating the risk you pose to others, don't subtract that off, because you could be in the incubation period and not know it.)

There's Figure 3 in this article in the NEJM. It's only a single study, and it concerns only the Pfizer vaccine. You might have seen a crude version of it in this xkcd cartoon. Note that the y-axis is cumulative incidence of Covid-19; to get a measure of its effectiveness at a particular time you should really look at the gradients.

You can find fat powerpoints by the companies on the FDA's website: J&J, Pfizer, Moderna. Each deck includes a Kaplan-Meier survival curve, showing COVID protection by day-after-vaccine. On the J&J deck it's slide CO-36. The curves for the placebo group and vaccine group get further and further apart as time goes on, with the lines first separating at the 14-day point. I am not sure if this means the vaccinated gain additional protection in week 3, 4, etc., after vaccination, or if it just means that more placebo people get infected as time goes on. What's interesting with the J&J plot is that the lines don't visibly separate until day 14 - naively this makes me want to say the vaccine isn't effective at all till day 14, but biologically that doesn't make sense to me, so I'm not saying that.

A potentially more interesting question is what vaccine effectiveness as a function of time looks like near its protection expiration date. Not much data to answer this one yet. May become relevant in 6 months in a worst case.