jjsakurai

TCS is basically math (although less background in math topics is required to get started than most other math fields). If you're not good at math, then you won't be able to do research in the area.

If Finaid is important to you, then it's very tough. Most US universities don't have finaid for MS students (and AFAIK, no US university admits MS students with any guarantee of financial aid - infact you usually have to prove that you've sufficient funds to cover the cost of education, either through your personal funds or loans). It is possible at several places to get a research assistantship (RA) or teaching assistantship (TA) but only a small minority of students get that so it's not something I'd recommend you rely on. Policies at every university may differ so I'd recommend contacting MS students at universities you're interested in to find out how common it is at that place to get an RA or a TA. Toronto is a top university - comparable to a top 10 university in the US so admission is hard. Try and apply to the other Canadian Universities I mentioned. Some other good places are Simon Fraser, Carleton and Queens University. Regarding transfer - I've never heard of an MS student transferring to another university. I guess it's possible but you'll likely end up paying even more depending on how many course credits you can transfer. Also transfer admission is usually a lot harder than regular admission.

My guess is that you've a good shot for a Masters at the top 30 or even the top 20. I'm not terribly familiar with MS decisions so it's only a guess. I'd also recommend that you add Canadian institutions to your list (Toronto, Waterloo, UBC, McGill, Alberta). The tuition is usually a lot cheaper and it's somewhat easier to get financial aid.

My best guess would be that it might be possible to get in a PhD program ranked say 75-150 based on your background. Emphasize your CS related skills. The statistics stuff you've done should be helpful. Definitely take the GRE again and get a good score on the Math section. If possible, talk to the CS profs at your current university for advice. I'm just guessing here though and don't have any concrete knowledge.

Since your goal is to eventually teach CS at a college level, I'm not sure whether a PhD would be worthwhile. A PhD is basically a degree in research. Unfortunately I'm not familiar with the best way to become competitive for a teaching job (though I believe several community colleges are fine if you just have an MS and not a PhD) If you do decide to go for a PhD though, then I've the following recommendations - 1) Taking the GRE again and getting 780+ on the Math section 2) Applying for a Masters (ideally atleast a 2-year program) and doing research while pursuing it. Alternatively you could also try and start out as a research programmer at a University lab and then gradually start doing some research too. I still feel you'll need a Masters though.

There's also scholar.google.com (though that sometimes shows pubs from a different person with the same name) , DBLP (it's not totally complete though) and the person's own web page. My impression is that at the top 10-20 schools, professors usually have 1-2 pubs in the top conferences of their field every year. For the really highly ranked schools (say top 5), profs usually have several pubs in top journals and conferences every year. Be cautious though - AAAI/IJCAI are not the best conferences for every field. Many subfields of AI have their own conferences which they consider the best. For instance, ML has NIPS/ICML as I mentioned above, game theory has AAMAS/EC, etc.

While I'm not familiar with RPI, one way of figuring this out might be to look at the publication lists of potential advisors. Do they regularly publish in top conferences and/or journals of their field? If so, then the research is probably competitive. For AI in general, the top conferences are AAAI & IJCAI. For ML, it's NIPS and ICML. Don't know about Cogsci.

Most of those people have done either BS or MS at well-known schools...

And how is the adcom supposed to gauge your preparation if they're not familiar with the courses and standards of your university? Citizenship makes little difference to your chances (apart from the fact that you can't apply to citizens-only fellowships which can indirectly hurt your chances) While it's definitely not flattering, I wouldn't call it detrimental.

Every researcher in the field would know the various conference/journal tiers. Unless your advisor has a vested interest in a particular conference/journal, it's perfectly safe to trust your advisor's opinion on this.

Publications don't matter that much. What matters is recs from professors that the admission committee will know about (IIT Kanpur should have quite a few of them). Also, CS Theory is driven mostly by conferences (STOC/FOCS/SODA/ICALP/etc.) rather than journals. I'd take that option if I were in your shoes. Also try to work with Manindra Agarwal if you can. He's a big name in the field and a rec from him would help your case a lot. It might be worth it to apply this year and if doesn't work out, then do the RA and apply again next year. Talk to your profs who've experience with students applying to US PhD programs to see what they think. The latter is perfectly fine. Publication count doesn't matter too much. Depends on the details of your applications (recs & research mostly) but Top 5 from IITK is definitely doable. Even MIT/Berkeley/Princeton are doable depending on the quality of your research and recs. Do good research with various professors. Ideally you want to have as many research recs as possible.

Given that you have a 3.6 for your Masters and come with external funding (I assume it covers both tuition & stipend?), I'd say you definitely have a chance at Columbia/Yale/USC. I still feel the rest are out of reach but I might be wrong. Given that you did your Master's at Harvard, you might have a chance there if the professors have a good impression of you.

I'm not saying that it will give you much of an advantage. What I'm saying is that if you come from a school that no one in the adcom knows about, that will hurt your chances as the adcom will find it harder to gauge your preparation for grad school.

Yes it does. Berkeley is known for having a very good CS curriculum so your GPA is viewed in a somewhat more positive light. The profs you're getting recs from are much much more likely to be known by the Adcom which is very valuable. While a degree from Berkeley may not have much of an edge over a degree from a place like U. of Michigan for PhD admissions, it's definitely a lot more valuable than a degree from a school no one in the Adcom has heard of.

A PhD is usually a paid position. You don't need to worry about funding. To be completely honest, I'd say that given your lack of research experience and your GPA, your chances at the schools you mentioned are close to zero. I'd recommend getting some research experience before applying and add some lower ranked schools to your list. Even with research experience admission is going to be tricky given your GPA. You may wanna take the CS GRE to make up a bit for your GPA. A good way to get research experience might be to say try and join a university lab. If you do good work there, then you can get good recs and the prof might even be interested in funding you for a PhD at that university.