* Intro to AI I'm not familiar with Udacity's new data science courses, but I think Udacity has the best course presentation generally.
* The NLTK book (http://nltk.org/book/) has a lot if good info, even if you're not doing NLP.
* scikit-learn also has some good machine learning tutorials
* A Programmers Guide to Datamining (http://guidetodatamining.com/) has some good practical advice and is very easy to read.
It's hard to recommend what will really help someone without knowing their background.
Complete projects or enter Kaggle competitions - https://www.kaggle.com/
Learn as you go like everything else. Once you've finished a few projects with good results you should be able to land a job in your chosen specialization.
curious_dream said
Coursera and Udacity! You may have to wait for a couple of years to get all the courses you want, in, but, on Coursera there's Andrew Ng's famous ML course, and, recently, a few courses that use stats and 'R'. 'Data Science' from last spring taught some Hadoop, and Udacity has an open course on it right now (no set schedule). And, Kaggle, as the others point out.
I'd say it really depends on what 0 really means for you. "0" for me was a physics degree and handful of years in industry doing physics-y things. "0" for somebody else might be a stats background, a programming background, maybe even a humanities background. The advice is going to vary vastly depending on the context - and it's going to vary a lot based on location too.
Here in the Bay Area, the demand for data science people is a lot higher, so there are a wider variety of paths into the field. Try to break in on the East coast, and it might be much harder.