DBTB INT Michelle Casbon

Michelle cast on independent senior Dana science engineer the most exciting thing is talking about idml which is a persistence layer for it has some core machine learning libraries in it and I was able to show the github repo link today unveil it and talk about the details of it I've been working on it for a long time since the end of last year and it was really nice to go into detail about everything that's included in the package well it's everywhere every piece of software relies on it it's it's sort of at the core of anything that anything interesting that you want to do and we have so many more tools that we can use to look at it to explore it to find interesting things about it and there's so many data sets that if you combine them in different ways things that you wouldn't expect to combine together sometimes you can come up with very interesting insights you can do a lot more sophisticated things with data then you could before in a lot smaller amount of time that's exciting you don't have to reinvent the wheel there are a lot of machine learning libraries out there and there are a lot of NLP libraries out there but it's still hard to operationalize the NLP process and I think that the software that I worked on does a lot of that and does it in some really unique and interesting ways so I think I guess the core insight would just be everyone's kind of working on the same problems and so if we share what we're working on if we share how we're doing things that how we're solving those problems we could all come up with better products you okay your hands dirty just start playing with things find the data set that you think is interesting or come up with a hypothesis about a data set and just start exploring go through tutorials do really simple things you don't have to do anything very sophisticated talk to talk to other people in the field talk to me talk to anyone at this conference I think that it's really key to have friends in the space and people that you can bounce ideas off of and generally people are very open to talking about what they're most passionate about which is their everyday work so i would say for me becoming a really good one has been not just the practice and not just learning on my own but those relationships and talking about things with my friends with other data scientists with people even in fields that are sort of surrounding data science because they have a lot of exposure to other data scientists and they can they can come up with they can just make sort of offhanded comments that that will bring about for you like oh I should really try that technique i think that algorithm would fit in really well with the problem that i was trying to solve so get your hands dirty and make friends in your industry you