Scale By The Bay 2019: Erik Repple Interview
Recording: Scale By The Bay 2019: Erik Repple Interview
[Music] my name is Eric ruble I'm a machine learning platform engineer at quinby's [Music] it's yeah it's my first time speaking at scale by the bay it's my first time speaking at any conference I heard about it because of LexA reached out to me on Twitter I've heard of scale by the baby before just you know it's a famous conference but yeah that's the first time I ever thought about speaking out it was like Syria [Music] yeah I think my favorite part about my talk is that it's like an actual narrative about a thing I tried to do and so it's like it's not some kind of like kind of stuffy lecture about like oh this is the it's it's actually like I live-tweeted a project I tried to build and then you're like walks you through like my progress on the day and then the most important part I think is like the takeaways and kind of like the learnings of like here's kind of the where we are in terms of machine learning on mobile and here's where we could get to in a few years that everything goes well it's mostly coin noises machine learning is around fraud we have some stuff around like the UX and kind of improving the quality of platform for customers and I think that that's only going to become an increasing portion of how we deploy machine learning and machine learning mobile helps our company scale because it means like it reduces the amount of compute resources we need to deploy because all the models can be just run on mobile effectively customers phones are like we're all the model evaluation which we would otherwise have to host is actually run [Music] I would actually I'm really looking forward to continuing to see engineering effort go into machine learning I think that machine learning is often treated as mostly a research field and we haven't like we're only starting to scratch the surface as to what how we can make it better aligned with engineering as a busy as a domain yeah I'm really looking forward to pages Swift for tensile flow doc that's probably the one I'm most looking forward to and what else I was just gonna flink talk that was pretty good I really liked I forgot the speaker's name I really enjoyed the were we right all along the talk about Moore's law and how rights law may be actually the better better indicator of future progress [Music] at skillet it's been really interesting to see the like breath of people who are here and companies and what they're working on and kind of the how everyone is roughly trying to do the same things but they're kind of taking slightly different approaches and everyone's kind of learning from each other I think that skill by the bay it's a it's only like the second third year that it's been scale by the bay rather than Scala by the bay I think that it's still very like Scala heavy and I'm looking forward to now that like Pratt like cutting edge deep learning is mostly Python and we're moving kind of into a hybrid world where not everything is Scala I'm looking forward to like kind of more like more non Scala based talks you can definitely feel like the Scala roots and as a guy doesn't really program Scala it's like very admirable but I like it's yeah it's not as much of a thing that I'm into [Music]