DBTB INT Alan Ngai r

[Music] my name is Alan I I am the CTO and co-founder of absurdity I think it's for me since we're a monitoring company is not something that you would typically think as a you know when people think about data and data applications they think data scientist big data processing information and the management aspect so that people don't immediately think right and it's really interesting to see you know that the talk I gave that the engagement and the number of people who actually care about this stuff even if your data scientist because typically datacenters you know they don't touch operations as much but you know take care about data so something in the infrastructure or in the management system is affecting your data and affecting your results you care about it and and so you know it's it's really nice to see that you know people recognize the problem of all kind of just not just the data but actually the infrastructure supporting that data so that's good validation for us I think it's just kind of the next logical progression of what you can do with technology i mean you know prior to this it's it's going to the cloud name for architecture you know scalability performance and whatnot but i think more and more people recognize that the next set of values comes from data you know from fraud detection to personalization to search all of these are data centric and you know you know businesses you can go into business just optimizing for that so their business out data helps you basically for example manage how you how frequently talk to customers or whether customers will churn and that by itself is completely data centric but that drives huge business value for you know customers so I think more and more people are recognizing this the ecosystem and technology is available to actually make use allow people to actually put stuff together to take advantage of this where I say ten years ago you had to build everything in the house and you have to have an army of people today a small team of five engineers can put something together to solve a problem using data so I think you know the time has come for everything to come together and for us to you know [Music] I think two things number one organize your data and so the talk was about a monitoring and I know how to get monitor and data applications and I think the number one note that I would reinforce the scene organize your data organize your data by concerns you have a data pipeline you care about throughput care about error you care about latency regardless of your business concerns so you want to be able to take a look at those metrics you know together put all your data in one place so that you can actually you know cross-correlate these things are by their very nature distributed so if you have five different tools for five different pieces of your pipeline it's impossible to monitor so now get everything in one place organize your data is number one number two you don't have to do it alone you know their companies like us that are make the problem a lot easier so you can actually solve your data problems and you know rely on you know somebody else to help you with your monitoring needs [Music] I think probably the biggest key point is to actually solve problem it's going to sound trite but I mean there are coarser courses and you know you know online courses that helps you get started on on you know machine learning and I don't like someone not but I think the most important thing I think as it's true with anything is to basically actually try to solve the problem get a data set figure out something you want to solve you know try to predict for example I was a good example so for example a netflix has a paper out on how they do cross validation you know how they take a set of users and they take a set of movies and predict which users will like what movie and they give you that and try to solve that problem right and you know just go through the process and then you know once you get used to that actually try to identify a problem that maybe it hasn't been solved and just to try to do it i think the most important thing is try to get real-world application and the key thing is data science is much more than just the techniques and the algorithms and the function is really getting an intuitive sense of the data so i think not only should you try to solve problem but try to solve a problem where you have intuition on right like you've got a background in I don't know ads you know solving a problem you have a background in you know search solve a search problem because a lot of it has to do with okay looking at data and trying to see the non-obvious features you can extract from that data and try to combine those features in various ways to solve problems so without that intuition you can actually do it so yeah solve problems and go with your intuition and within a domain that you that you have expertise in you [Music]