This thread is of particular interest to me, as I also have an admit from Michigan's MA program and I also am interested in getting some (or a lot of) Machine Learning exposure. I've been leaning towards UCLA over UMichigan because it seems to be a bit more flexible and potentially more rigorous (but hell, what do I know?). That being said, Michigan seems to be a much higher-ranked and higher-regarded program. Anyway, here are my impressions of the Michigan MA Applied Stats program at the moment (for what they're worth):
I researched ~16 randomly-selected alumni from the program that have graduated in recent years. I was able to find information on all but 2 of them. Of the 14 remaining, 11 of them definitely seem to have had TA positions, so it does look like getting such a position is fairly likely if you try. Also, almost all of the 14 seem to have gotten positions following graduation, of which I made the following generalizations:
Most had titles like statistician or biostatistician
Most got jobs in the Michigan area
~2 went on to PhDs
~2 went into finance (ibanking in one case, risk officer in another case)
As near as I can tell (and, again, I basically *can't* tell), none are doing anything ML-related
For the program itself, I noticed that there are a few places to customize your degree outside of the regular coursework:
You get 3+ electives, all of which (as far as I can tell) need to be taken in the Biostatistics/Statistics department. These are pretty flexible, assuming you choose something in their pre-made list or something above course level 600+ (I assume higher numbers = more advanced classes).
From what I can tell, you are required to take 1 cognate course, which can be in any department, so long as you get it approved by the stats department. It looks like it's possible to fit in a second cognate course, but I'm not sure how difficult that is along with the rest of the degree requirements. This is where I'd think you'd be able to take a CS class or two in Machine Learning or a related topic (though there is one 600+ stats class that can be taken as an "elective" that touches on Machine Learning as part of the coursework).
To conclude, Michigan has a great reputation for Machine Learning, but it appeared to me at least that getting a significant amount of ML exposure would be difficult in this particular program given the low number of cognate courses.
Hopefully this was helpful. I'm planning to visit Michigan soon to hopefully find out more, and can potentially report back once I do that. Please let me know if any of the above looks wrong, as it's what I'm basing my decision on at the moment (and I don't want to base my decision on bad data!). I'd love to hear any other thoughts.