I don't know much about ML/AI - Where do I start?.

Sounds very familiar. If you are like me, who studied AI as one of the undergrad comp science subjects, I am sure you heard about it on and off through last few years...probably watched IBM Watson segment on Jeopardy or on 60 minutes.  

Meanwhile, life went on, building other software systems and products.  Then you heard about Big Data technologies and adoption in companies you work. You hilariously saw few people, who claimed to be Data Science Experts, Analytics Chiefs (too many chiefs right :-)) and started talking about machine learning, data science etc and big data..Lately AI....

So, you start taking more interest and say, yeah it makes sense to it now.. There are so many practical applications for it....You convinced that, finally there is huge volume of data, that can be gathered and processed.

That is how, I got interested again in this topic after college(where I thought AI was so cool).

So, I started reading some books(more later on some good ones)...I realized, I had forgotten lot of mathematical/statistical concepts, I should have remembered from college. So, started learning them...

Some useful tutorials are available on the web..If you have undergrad background in math, comp sc etc, you don't need to pay 2000$ for Andrew Ng,s course :-). You can self teach your self all these through a methodical way.

Once you brush up, the concepts, someone would have told you about Scala and comeback kid python(In my case, my kid is probably better than me in python programming).  So you take care of programming language aspect...

you would have also heard about Hadoop and several hundred big data platforms and companies. And ofourse one or many of Amazon's web services to do all this. So, you took care of that.

Then you are really starting with how do I process data in real time?. How do I write ML algorithms and implement them?.  How do I process large sums of real-time data and apply ML algorithms there?. How do I build systems that are "real" and can be deployed to productions...

I plan to discuss all these topics soon.

I might have made it all look simple and easy...It obviously isn't and required spending lot of time...But definitely self-learnable through concepts you already know...

Some of good urls are...(Will update with more soon)

http://deeplearning.stanford.edu/tutorial/