data.bythebay.io: Justin Basilico interview
Recording: data.bythebay.io: Justin Basilico interview
Justin Basilico I'm a research and engineering manager for Paige algorithms engineering at Netflix what was the most exciting thing for you about being and speaking that dated by the date today I was happy to be able to talk at a place where there's a lot of people who share a lot of interest in machine learning and doing data science and solving you know really interesting problems why I think data is cool exciting is because I think it really lets us solve new hard problems that we aren't able to solve using other techniques and so I think when you combine really good high quality data with good you know intuition and good algorithms that you can get really amazing results and solve problems that otherwise people might have thought were impossible so made in sight from my talk was that when you're thinking about building machine learning and software and putting into production systems that it's really good to think of the whole problem holistically and that at the end of the day machine learning is really about you know doing something in a real application and actually taking actions based on the data that you're using so looking at the application actually do you know kind of thing about developing that to help make machine learning easier and making it easier to rapidly iterate on ideas and experiment with things and make it very flexible and modular so that in the long run you can really test ideas quickly and make sure that you know there's no discrepancies between kind of your lab environment in your production environment means that you'll be able to find you know you really come up with better outcomes in the long run and you'll find things quicker and you'll just you know make you lick the most that you can out of you know putting machine learning in your system and make it as good as possible how to become a data scientist I think there's obviously a lot of paths to it because it's a kind of a multidisciplinary area so I think you know whatever path you're on you're probably just going to have to learn to build up a lot of skills that may not come out of your area so if you're coming at it from you know computer science you might need to get a lot more deep into say statistics and things that you might not otherwise learn for coming more from you know maybe some of the sciences being comfortable with coding so really think of them about filling in becoming a very broad building up a basically very broad set of skills so you can effectively operate in those areas and then also working on kind of also how you communicate with people and on the human side because end up having to interact with Al and append on a lot of other people if you're doing data science to actually be really good at it and then you know to kind of once you have the basics to get really good i think really just about getting your hands dirty and trying out lots of different types of problems and you know and you know trying out lots of different types of algorithms and you and then deep diving into the area you're working on and trying to build up your own knowledge and intuition for the data and the problem area because ultimately you know as the human involved you it's on you to kind of figure out how to come up with great solutions for the problems you're trying to tackle you