I am a software engineer who made the transition to data scientist. It's not so much about learning an additional language like R or Python, but rather about acquiring knowledge/skills in experimental design, hypothesis testing, statistical analysis, and machine learning. In other words, you need to learn how to think and approach research questions like a scientist, and have sufficient knowledge of tools and techniques to answer those questions. I learned this stuff enroute to a PhD in linguistics with a specialization in cognitive neuroscience, but they are plenty of resources available these days that you could learn much on your own through moocs and reading.
I think the acquiring of the knowledge/skills mentioned by Brandon is most significant than others, that have the role a support for the firsts. Besides, the article wrote by Vicent GrandVille helps to define the many ways that each professional can follow based on the goals and the characteristics of your work market. I think that, in work markets like Brazil, you have low penetration of professionals with large academic experience and academic profile and therefore our work market tend to have a greater demand for technical skills.
Thank for your response Brandon
I'm a software developer and my objective is to make the transition to data science...I'm studying currently data mining and machine learning..
I hope that my experience as a developer will be a plus when combined with data science knowledge
I would suggest you to first focus on these three topics: introductory course on statistics, data visualization, and machine learning. There are courses on these topics on Coursera, I suggest you to start there (before buying any books, or signing up to any classes). Data Science is mainly mix of CS and statistical majors and I still haven't found a good academic program which covers the whole topics. Stanford and UC Berkley have some 2-year programs, they are super expensive...Basically, data science contains a very wide range of skills, and it is quite difficult to cover all the topics with a fair amount of attention in a 2-year program, that my belief. Anyhow, other than coding skills, if you start studying on those topics, and practice on solving data science problems, you will gain knowledge and experience of working with data gradually. Since you are a developer, learning R or Python would be matter of few days for you, so I wouldn't suggest you to focus on that...
I suppose it all depends on what type of Data Scientist the person wants to be. On the one hand there are professional data and business-savvy statisticians, and on the other extreme, junior data plumbers with a passing acquaintance with equations. The odd things is that Data Scientist is another one of these terms that will be eventually trivialised into the zone of meaningless. Oh, the joys of the world of Big Data and postmodernity.
I'd say the hardest part is already behind you - becoming developer! All too often I see "pure data scientist", i.e. without software development background, stuck when things go south. With increasingly more complex tools and frameworks, good understanding of software machinery behind it becomes more and more necessity. You also have an edge when it comes to usual day to day business: compiling a library for *nix, deploying a database, building front-end to it, integration and testing. It's easier for you to provide a complete solution, not a piece of it.
Others have already provided some valuable advices, so I am not going to repeat it. MOOCs are cool, but there is nothing better than real-world experience. If you cannot get it on your current position, play Kaggle or, even better, find the cause you care for and start coding. There are plethora of non-profit organisations that could use help of people with your skills.