sfspark.org @Trulia: Alexy Khrabrov interviews Hesam Salehian and Chul Lee of MyFitnessPal
hello everybody I'm Alexa crab Roe fourth organized and forceps park and France made up and Hugh are on location at Trulia where we have an amazing meet up with my fitness pal folks are truly and asam's likey on who is the director of engineering and a senior scientist they use spark at me my fitness spell to hopefully get us all healthier and they're going to present later how to do this technologically but this is kind of to get to know them a little bit better I'm going to ask us on some issues so welcome guys it's good to thank you thank you I can you tell us a little bit about your own career paths and how you came to my fitness pal and what's exciting about using data analytics for for for health sure so my name hi everyone my name is Julie I'm a director of engineering at my fuse power and my team is responsible for pretty much everything led to data and prior to my friends pal I work at LinkedIn and a major team that was responsive for content relevance listen and then the reason why I joined my fins was because like my friends power is professional number one app in the category of health and fitness right and then we were forefront of that kind of innovation kind of new wave of disrupting health and fitness wheels so that was pretty exciting and then my fins by a lot of data so I'm dated I me as a imagine so I was very excited about you know what kind of stuff that I could have done through a microphone so i decided to join my facebook so school how long hello everybody my name is hossam i joined myfitnesspal in January 2015 before that I got my PhD from University of Florida my full of a study was more focused on computer vision machine learning and medical image analysis the reason the main encouragement for me to join myfitnesspal verse number one it was the best app in health and fitness and they had a lot of data that I could as a data scientist I could get involved and apply like several machine learning techniques that I learned a lot but and that was a very very great opportunity and another opportunity for me is to affect people's lives and to help them live healthier which was very important and I am really happy you think so maybe you know I'll continue so with the stool so I mean we'll use linked it right like we all get all the information a lot of updates from LinkedIn right kind of user of LinkedIn so how do you transfer this knowledge of social graph and content and pulse and all this textual data to my fitness pal day they can you talk a little bit about what kind of data is there and kind of what kind of techniques are transferable you know from one domain to the other you mean from social through social from LinkedIn took my fitness problem right so I think basically the the the whole idea is pretty much the same in the sense that okay so in the case of LinkedIn you collect a lot of you know profile data a lot of social graph data and so on and then out of the you create value for our customers right so i think my fins pie is pretty much trying to the same in the sense that we are we have collected a lot of data is pressure on nutrition data ok so we're have collected really like pick this knowledge base of food items and their recipe items on right and then also like a diary information from our users and once that we do that we try to create value for our consumers or four for our users right so that means like we want to bring you more insight right we want to help you with planning right and then we want to make you healthier like and then we want to make sure that you can track so all these like activities and value prop for for our users that's what we're trying to do ok is there any social graph date associated with this there's a little bit of social graph uh not that much because like this nutrition tracking tend to be very personal right now is definitely up to you whether you want to share the data with your friends or not right so where does social yes but I say is new kind of social right so so it's a little bit different from a traditional sense of social yes right so is tends to be way more private and they're more precise it's harder for collaborative filtering like you know I have a friend who isn't beauty business and he says you know he's selling essentially cosmetics through kind of question answer system and I think the inside was the partially a very interesting company rights only the inside is you know you cannot do this like Amazon because you know if your friend has red hair and you have been alone here a different kind of skin you cannot recommend beauty products to each other because I can be different so does this apply to 2 Timothy to spell right yeah that's dusty kind of the irrational right is there anything to deliver because there are any similarities you can explore you know because essentially like what works and fitness you know like willpower hopefully shared value right like I mean other things like this you can compare for one person and recommend to the other person yeah that's def a like you know possible I like you know it so social media companies they have really you know you know innovated that you know aspect of a like you know somebody else knowledge it can be reused or we collected right for somebody's like knowledge so from that perspective my fuse body is actually trying to the same in the sense that the log data or like recipe data that has been generated by somebody else is we collected and then reused again for other users right so from that perspective like this aspect of crowdsourcing of data and then creating more value out of data that other people have used from that perspective there's a seminary right but in a but but at the same time very different again went from the traditional sense of social okay because yes is social but it's not like you know very kind of active social network yes because it tends to be way more personal way more private so it's social but different kind of social okay cuz I'm what's your take Alice I am to add some stuff to what you'll mention we are definitely looking for some like patterns for the food consumption or the users or even more like tech exercises when we have joined now under armour and MapMyFitness community so definitely finding some patterns and providing like relevant recommendations based on the user's history that's something that we definitely love to do and it's pretty exciting for us and the thing that is that we have lots of data and we have a huge set of history for over like millions of users helped us a lot to give more and more accurate SMS for recommendations and that would be like the more focus for us ok I couldn't took a little bit about kind of the same kind of knowledge transfer between the area we did image and vision to my fitness pal data sure um did the main type of data that we've been dealing with were nutrition information we haven't had a chance to go through like the image data yet but since my background was more biased towards machine learning and a specific example was computer vision I was really glad to work on machine learning algorithm applied on nutrition information the main challenge for us was to handle like a large size of the data set and that was one of the significant reasons that we use the spark for a data processing and that was really convenient to use spark for machine learning purposes and we've done several machine learning projects on our nutrition data and our food consumption data using a spark and that was a fantastic experience okay Oh interesting so is it possible like you know just hypothesizing like it so you have let's say a dietary requirements you need to control your do ahead is it possible to take pictures of food and figure out caloric content of food yeah that's a very long term vision but I mean that would be really exciting to work us to work on and we definitely like be interested in working on that but that's a long term yes you know I've been when I was in grad school I was kind of thinking you know I found just came out and I thought even thought like you know I was thinking for start-up ideas even thought up in our name each shot like you should a picture you know what you eat in kind of but apparently it's hard because you need to figure out what's all the ingredients that's a very challenging problem for yeah yeah yeah yeah so they're like there be two types of challenges right want to be technical my summation they just stopped by themselves it's really hard to solve right and the other challenge that that we have to overcome would be how is that product experience going to look like well yes because okay so how is that flow is going to look like and what's going to be the kind of the value that we can create for customers right so when to take you know photos or pictures right right probably you want to take the picture and say the later and log later right so there are two types of you know challenges that we have to overcome but yeah we're definitely thinking about like those directions cool yeah I want ask about another aspect of kind of using technology to better our lives right so I was just you know a few conferences in Europe and one of the most deploring summit in London and they had a company several communist you'll present this idea so their personal assistants a very popular right now so so they have a company which kind of talked about you know using deploring for acrylic and all kind of inputs in our lives and kind of helping us leave you know better lives have no complete attention right but it's kind of it's it's tricky because you know we by using another app to help us with overwhelming information which is all coming from other apps right so how how did I see this playing out you know like if users kind of need to focus on improving their lives right and they need to step away from technology then you to exercise they need to change their diet I need to meditate and then we do something different right so you know but we're in technology so we need to monitor this like we need to have a net how do we solve this kind of paradox right we need people to kind of this is the Sun from technology or do make technology seamless how do you see the future of this that's a trick question to answer actually uh but i would definitely think that using technology users are getting more and more satisfied that was the experience that I have had in the past like adding a new feature to the app has a very very interesting feedback from users and they would enjoy like having new features and I think when we have like all different source of information from users like food nutrition exercises and things like that that are pretty much related to each other but from different sources that would be very exciting for us to work on like making really relevant recommendations more and more accurate for the users and that's I mean I'm pretty excited about that and I'm sure users would be excited too yeah so I think you're really hitting a point around like I've really like you know this whole experience around tracking and then logging is okay and i think my physical has done pretty soon job right but it's not you know that gray yet way because because of the reason that you mention is not very attractive right is not very kind of human friendly yet right so this is I think why also like the deed you know the coming revolution in AI good knob to right so as the AI technologies get better right and then all these apps will be more intelligent and more human acts and therefore like the the friction around logging and tracking will be eventually removed or diminished right and then in that way like this whole app and the tracking experience will be more like human eyes and hopefully like you know once I happens and we can be part of that way oh and so I kind of wrap up with one question do you know for each of you I you know do you get to interact with your users and like what's the most satisfying part kind of using technology to better human life do do you see any user stories do you see any success stories like what's what's most inspiring for you so yeah we have we're strongly based on the user's and the users feedback and we have a channel that we share the feedback from users especially very very impressive the stories for the users that have lost a lot of weights during the last and actually we have the pictures before after pictures from users on the pillows on the couch and in the office and also very like famous quotes from users that are mentioned on the wall in the meeting rooms that is really really inspiring for us and that's that's the number one reason that we get to work every day and we try to make our best that's a tricky question the one thing that they really are catches my mind but I'm not sure if it's from users but it's like being moderate is awesome especially in being madre so that's something that they can remember interesting well since you mentioned this I cannot help but ask you know what works and losing weight somebody wants to lose away you guys are data-driven you you don't use the here so you have real dating what actually works he was this one little weird trick to lose weight use my fins puff okay then you ready to like it this way doesn't anything yeah yeah yes awesome thank you guys that's great we're looking forward to at all alright thank you for unity thanks thank you