Scale By The Bay 2018: Chris Fregly Interview
Recording: Scale By The Bay 2018: Chris Fregly Interview
so my name is Chris fregley I am the founder and CEO at pipeline AI here in San Francisco for focused on real-time machine learning and continuous training continuous model improvement throughout the lifecycle of a model after they've been moved into production yeah the streaming parts obviously the coolest thing the biggest differentiator is the ability to continually improve models online most people when you think of a machine learning company you think of offline batch SPARC workloads distributed tensorflow offline but we're basically bringing these pipelines live and in production so bringing forth a lot of my old Netflix experience and my day to Brick's experience and combining those into one platform yeah and pipelines my company is very much on the edge of application development so software development as well as data pipelines and machine learning so really I've been bouncing in between those two so best practices around the infrastructure and sort of intelligent like infrastructure ways to scale out your machine learning but not too aggressively ways to scale it down you know yeah just a lot of best practices from these internal systems you see at uber Netflix you know Google these kind of places yeah there's a lot of good talks there's a couple good talks on machine learning in production so I'm gonna be paying attention to those there's some like newer open-source projects that are sort of end to end machine learning that I'll be paying attention to so yeah yeah a lot of good stuff coming out of Google in those folks these days you know I'm a traditional Java guy from back in the day so when I learned Scala it was it was pretty magical there's you know quite a lot of good Scala frameworks as well - quite honestly since I've been doing more with machine learning I've been doing a lot of Python and now that I'm you know we're kind of coordinating the UI for our application we're doing quite a bit of nodejs and javascript so you see functional principles throughout all these different languages but yeah just the readability the crispness of you know the constructs the basic constructs the concurrency constructs things like this let me see I love that it's right down the street from where I live so I could pop in and out and I'm on my way to the gym right now so popped in for this interview and then a quick panel later today but yeah it's a lot of the you know familiar faces so you know a lot of people from all parts of my life from my Netflix life from the cassandra' world you know all the way to my former roommate I think is sitting right behind you right there that works for Google so yeah people everywhere [Music] you