How and what to self-study Mathematics and Statistics to become a professional 'Statistician'? I already have a Bachelors degree in CS.

I have been trying to self-study the mathematical fundamentals for machine learning for some years now, and the requirements are pretty similar I think, so here you go: To study almost any other mathematical topic you need to know a fair amount of calculus, including multi-variable calculus and topics like optimization, so its best to start with that, or you will get stuck over and over again learning other topics because of poor calculus skills. Since studying calculus gets pretty dry at times, you can study discrete probability meanwhile, which does not require calculus and which will give you some of the most basic background for statistics. You will need some linear algebra both for multi-variable calculus, and for practical statistics as well. It isn't an obstacle to understanding other subjects as much as calculus is, and you can probably learn it as you need it, but it is anyway best to study it separately if you have the time. You need to know that many subjects in probability and statistics come in two editions - elementary, calculus-based one and advanced, measure-theoretic one. Also there are two approaches to statistics itself, the frequentist one and Bayesian. So, many subjects can be studied several times from different angles. Once you know calculus and linear algebra, it makes sense to study calculus-based probability theory, from a good textbook covering both discrete and continuous probability spaces. You can follow it with elementary, calculus-based  texs in statistics, stochastic processes and statistical models. You might also want to learn basic Bayesian statistics, which is different from what is typically taught in a typical basic (frequentist) statistics course. If you have the above basic background already, there are two paths you can take - you can study more advanced, but practical, subjects in statistics and applied mathematics that don't require more advanced analysis - various flavours of regression, linear programming, statistical software, decision theory etc. If you want to go deeper into the theory, the next step is to learn real analysis, and then measure theory. With this you can learn measure-theoretic probability and then finally measure-theoretic mathematical statistics, which is where most of the theoretical research into statistics happens. Of course nothing, except for the time this all takes, stops you from combining both paths somehow. For many basic topics, you will find more than enough resources on MIT OCW: http://ocw.mit.edu/. Look for course pages for the subjects mentioned using Google, you can find good textbooks this way, sometimes solved homeworks that are very helpful for self-study etc. Good books for any of those topics you can find here: http://www.cargalmathbooks.com/ The American Statistical Association has guidelines for undergraduate curricula, you can use it too see what people get to learn in a typical degree program: http://www.amstat.org/education/curriculumguidelines.cfm