Talks # 5: Parul Pandey: Data Science, Diversity and Kaggle

Key Takeaways

hello everyone and welcome to the new episode of talks and in this one I have someone very exciting it spiral she is the first Indian Grandmaster in the notebooks category in first Indian women and she is going to talk about a very important topic today and it's a little bit different from the usual talks that we have seen but it's it's very interesting and it's something that we should care about so she's going to talk about data science diversity and cattle and before before I hand it over to her I would also like to mention if you want to be a part of this talk series and if you want to speak then you can find the link to the forum in description box so I just fill out that form and I will get in touch with you and now over to you parallel Thank You Abby shake it so it's a pleasure to be here on your show and I'm glad that you're discussing such an important topic which is like talked about sometimes but I think it's a very important topic as important as model training or machine learning and machine learning interpretability etc so hi everybody my name is Beryl and I'll be discussing on the topic of data science diversity and Carol and be running through a presentation where I talk about how these three pillars actually have a very defining role in my life how I came to join Kaggle my current company and what I am doing for diversity and what I expect and you people will take away from this session now just like Abhishek mentioned this is what a different slightly different presentation and for the ones who've come here for Gordon stuff and trainings small training they be really disappointed but it is even equally it is even more and for them than the others because I think we all need to understand today's world in what's happening around how diversity plays a very important life important role not only in our life but in whatever we do so this is just a little bit about myself currently I am working as a data science evangelist at each to dot a I it's a start-up based in Mountain View California academically I am an electrical engineer and I graduated in electrical engineering and after that I joined Tata Power and I worked there for quite some time as a lead analyst in the Power Distribution sector and more recently I have been involved with women in machine learning and data science chapter and I founded the Hyderabad chapter I'm also Catalan and very active and casual and I recently became cattle Colonels Grandmaster I'm also very super busy mom and like all the other moms I try to juggle my life between Kegel and data science and work and with minimal DSP type events and I'm also a big big fan of Sherlock Holmes if you could say I'm a big fan of Sir Arthur Conan Doyle so I think then I find that Sherlock Holmes is kind of a very I think very great data scientist if I could say so surveyed all most of his stories and I try to implement some other things in my a still entry data analysis and gaggle so coming back to this presentation this presentation is actually wired into three important aspects just like the name of the data science Kaggle and diversity and how all of these things are interconnected and how all these three these three things have their own importance in this data science field so when I talk about data science I will be talking from my own perspective and from my own learnings for me currently data science comes to me as a part of each to dot a I where I'm working and I want to talk about h2 in this presentation why because I think a lot of people still have a lot of confusion about what is the difference between h2o dot e I and H 2 O 3 and what is driver less AI so I think it's important to tell you what I'm currently working on and how am I so ciated my company so we at h2o dot a I mean Li we are working on two products one of them is open source which is H 2 O 3 so that is the name of the product and the other product is called travel sei and that's an automated machine learning platform h2o is an open source machine learning platform and has a module of auto ml so that's a how it is differentiated but tribal SEI is an automatic machine learning platform and while telling about h2o in this presentation I'll just come to that a little later there's some features of each to that I want to tell you like it has all the algorithms leading algorithms that we use machine learning you have access from different languages as an auto am component has a distributed in memory processing and you can even deploy it easily so I think deployment is one of the most important aspects of any machine learning model that you do that once you model is created once you've trained it you need to also deployed so these are the key features of h2o now that is the open source platform the next platform or the product is tribal SEI and this is how the UI of driverless AI looks like so next time when you see such a similar screen you should know that you know this is what exactly's right let's say that we should talk about and again it has all these important features why I'm telling you about this is that you've recently launched our Learning Center and that Learning Center essentially has been created to provide free trainings on fundamentals of AI and machine learning and along with that we have instructor-led courses on tribal SEI and h2o in all our other different products this is important here to mention that if you're a student or you're in academics so you are a professor and you are working on some problem you can avail an academic license of art products for free and you can work by using our platform so this means if you are also nonprofit organization you're an NGO what you are doing - good just feel free to apply for the kind of it license reach out to me or anybody it will provide you the license for a year and you can start working on your machine learning projects so if you are working and you know you require some state of the art you have requirements your college you cannot afford it don't worry you come to us and we'll help you in getting your machine learning project started so why I want to tell you here is that this is one of the important aspects of use a diversity or democratization where we ensure that all of you if you have want to work on you really have a problem you want to work on that but you don't have resources we are there to help you and this big forms one of the most important and vital pillars of actually data science making it accessible to everyone and not just a chosen few so if you are interested you can just contact me through LinkedIn or Twitter or anybody else who's on the hto dot AIT you now let's move on to the next paradigm and this aspect of my presentation which is very close to my heart and which is something that I think everybody should care about that's diversity so when I came to H to Dottie I I met Aaron Eden is one of my colleague who works at h2o and she is the person who spearheads the issue or to amend a program she is the founder of women and machine learning and data science sort of organization and she's also the founder of odd ladies also so women M LDS they have a mission and they have a very important mission that is just support and promote not only women but all other gender minorities and they've been working in I think we've been working globally across this continents and when I joined h2o there was no chapter in Hyderabad and then had volunteered to start one and it was in October 2009 t when we started it and today we have a good 800 class members and currently the meetups are going online so we have a good proportion of men also joining it and like now we are open to everybody but this one important thing that you make sure is that we get the speakers or the speakers who speak or females so this is a conscious decision and it has nothing to do with you know why only one M female speaker then why don't you the chance to men it's a very important because a lot of times I hear when people tell me that you know they cannot find female speakers or become and they tell me like would you like to speak at an event and I will just tell them if I have already spoken at five or six events and why do you want me to repeat every time should get new faces you should let other people speak and then this constant thing was we cannot find speakers and so have women presenting things we always share their details maybe they'll even handle or Twitter handle and that's not just to put it down into the YouTube you know description box it's because we put it there so that people who are looking for females because in contact them that way do you don't have to they will not come up with the same problem of you know we cannot find with speakers I see a lot of events happening I see on LinkedIn and when I scroll down a lot of events happen in colleges undergraduate students they were recognizing a lot of stab the stuffs happening in offices and one of the very sad part is that the hardly any female speakers on the panel I feel that this also goes out for students who organize a lot of stuff when a person organizes something he or she is if its leadership qualities and I know how it takes a lot of effort and energy to actually organize something so when you are a leader and you have exhibited those qualities you could also have to work even harder to find diversity in your panel so next time you organize something in your colleges or schools or maybe companies make sure that it has representation from various sectors and I don't only mean gender I mean geographical representation get people from different parts of the country if you cannot get from the world get people different opinions don't just get people who would always say good things about the event that you're than I think get people who have some other experience to share if I primarily talk about women and what you can do so for the people who are listening to this session today there's something which I would definitely ask you to help me in and help all of us who are trying to help to diversify the whole data science field you can see data science is a very vast field and you don't data scientists are not only limited to one particular problem or one particular group we have wide variety of problems that we're trying to solve through machine learning and it makes sense that we have representation from every sector of the society so if you are working in a company you should always understand that less women apply for position so you should make sure that you really like your job descriptions that way one of the reasons when we apply machine learning today especially NLP techniques is to gauge the hiding bias if you notice a lot of job descriptions today are written like this he he should be proficient in Python he should be proficient in deep learning that way so you can make sure if you are in HR or things like that or things are going in your company you should always advise them you know that instead of he/she make it gender neutral again if you are working in a company and if you're managing people and there are women who've been working under you you should be an advocates to them because not all the women will come up and raise their voice against something that's happening that's happening wrong and again even in meetings make sure everybody's heard don't talk just don't just keep talking and let others also speak the other thing that I was mentioning is about having a more representation general representation in your speaking engagements and conferences that you organize or meetups that you organize is be aware of manuals a Manor is a very interesting term and those of who you are not aware of those of you we're not heard of this Mandel actually stands for male panels so a panel which only comprises of main and it's a very common site actually a lot of it you see nowadays it's just one female and that too surely comes when people actually say you know what there's no diversity and unfortunately sometimes what happens is the female is just active in that position for female yes so it's like anybody will do we just want a female do you know somebody who can just come for this conference because you know we just need a few more for the sake of female that's not the way things should be done they're not as talented individual standing two females out there just make an effort to find them see other conferences that they're speaking try and reach out to them of course you will have to do something extra on your part but you can be doing a lot of good for the humanity mentoring sponsors those from the underrepresented groups try and support somebody if you can and again create a space that is safe and welcoming for all okay for instance if you are taking part in carrying competitions you are working on a team try it get somebody else on the team so that they just can experience how it is to work and how is to take part in the competition if you are going and see you are supposed to speak at some conference just make sure that it's not a manual and be an advocate so that and say I will not speak on this panel if it only comprises of female and if you think this is all this actually people are doing in inside one of my Connie he refused to be part of a panel because they had no female and nobody from the under representative representative group so it's not that it doesn't happen you just have to take a stand and once you start taking a stand things will definitely improve we had women and machine learning and he designs has a bad chapter we are going to hold a 7th meter at this time that's to Touro actually and hello humans because we've had speakers from caroling we have riches from resort tomorrow be having fun now cheaper to present on computer vision she's also CAD allure but the ones want to know so we have also found female speakers and then these females present their work there about ten other females who approached us and they tell us that they also want to present which is such a good thing so keeping that in mind we had organized your CFP that's called for proposals and I know a lot of men also had filled up here but am i sorry right now we only having three means but thank you again but what you can do if you're a male is you could encourage your colleague you can encourage your family member you would encourage your friend anybody you could share it and tell them to apply for it and we are not going to select anybody be able to give chance to everybody as and when like maybe in two weeks or we'll organize a meet-up and once in two weeks in that way so we are not going to choose on what is the topic you know so just encourage people they're giving a platform we want more females to come out there and and talk and present things so this is a very important this was a very important aspect of my life currently and diversity and getting more women into the field of data science and tagging and joggle why I actually worked so hard in candle and I became a grandmaster one of the important reasons was that I want you to second example actually because you cannot be an advocate for something which you do not practice yourself so if I had to get more females into Kegel I had to set an example and what a better way to set example than becoming a grandmaster so now when I achieve rate I think a lot of other females out there believe that you know it is achievable and yes it just needs little boys UN's persistence patience and hard work and but it's not something that you think you know it's not within our reach so with that I will just shift to the last aspect of my presentation which is gaggle when I started this presentation I said I'm a very big fan of Sir Arthur Conan Doyle's work and in one of his very iconic writings called the Hound of the Baskervilles there's a very wonderful code that is written and which is actually in the article said by Sherlock Holmes that the world is full of obvious things which nobody by any chance ever observes I think this is one most important aspect of what I've been trying to do on kaggle I started a Kangol I joined Carroll actually five years back but I never used it for anything else except downloading data and I wouldn't even work on that data in the kernels that have been provided by Carol I would download it and I loved the system and would work on that but something changed last year in 2019 in 2019 Carol hosted the Kaggle survey contest in that time I was searching for some data on women's representation in data science and luckily I found that data which had a lot of information about women gardeners of course the data was not a representation of the complete area science community but it did have a reflection and so I participated there and I just kept on working through the data and that notebook that I wrote actually became sort of a pillar or you could say what are the starting point for me so on cagin journey because that was so well received and not only was that well-received obviously it also got me a price and it got me a notebook price and and just suddenly things just you know dawn didn't mean that i really enjoy all this stuff and why don't I start getting active on candle more okay and but there's something again more important was this is that notebook actually and in this notebook I sort of created a story I so this okay so this notebook actually helped me to understand what is the how can you use data to present something and how you can use this data to actually tell people a story and how you can find important aspects from a data which is actually hidden and from there my kago journey started so essentially I joined carriage 2000 in a five years back but actually I started working on cable in October 2000 nineteen and I just got so engrossed and so addicted to it that finally june this year I achieved the status of blind master which I think I'm very proud of and I'm more proud because now I can set an example to others if I can do it in anybody can do it and there are actually people who are doing some wonderful stuff few of them are still in the undergraduate years I think and it's an impressive is the deep there's notebooks and I feel so proud about how people are using this platform for such good use I'm mostly active on notebooks because I have certain limitations with time currently but I did participate in the Wednesday death and computation of my colleagues and the part is but more because I wanted to know how it feels to work in a team for a competition now in four different people at four different places on the earth can work together and let me tell you I think team formation is one of the most under rated benefits of Calgary's when you create a team for any competition you learn so much from each other we learn maybe Bini that you learn and it essentially helps you to work together for this common goal and then distances don't matter their time slots won't matter the time zones are different but then you just come out and then you finally submit your solution I think it's great so a few things alone you're not able to get the mass amount of care and try teaming up with other people and this is typically my journey which starts there in October last year and ended in June 38 has an ending actually it's just the starting or something new maybe this year it's giving competitions and yes I think I'm going to enjoy that and again the environment in which you word helps you a lot in making you what you are I joined h2o in July last year I never i joined h2 one thing that i used to look up to word so the multi-talented people of course all of them are there this is the specific people there because the cattle grandmasters whom i would look up you know and see maybe like if i will be able to reach that place someday and today it feels very proud to have to even you know put on the slide next to them and i think you all can identify some of the very common names yeah this mod was Olivia this is marios again you've had the mystery here and this is SRK so the lie - Kumar is Rohan its Shivan and there's me and this Kim also came into competitions Grandmaster so we're very proud to be two women grandmasters on our team and this is a very iconic team and I'm so very proud of it and coming here has been made possible by a lot of work that have done but also by learning from others by learning from my mistakes from learning from the community and if you will so set a path for yourself I think things are achievable and just make the right use of every platform whether it is LinkedIn where this guy will but his Twitter and I think all of them have a lot and lots to offer so that was my presentation my life in the few slides and now I'm open to questions if you have any questions welcome thank you there are many questions I have selected a few of them and the questions still keep coming one of one of the funniest questions I got was what was your GPA and finally I graduated from an IT Havilah that's the NIT which is in Himachal Pradesh in India and I was a very good student yeah I was a very bad student but anyways yeah okay so yeah there are quite quite nice questions here and the presentation was very good if it was really fun to watch and also it's it's it's an important topic people should know about it and yeah one of the questions that I got was how do you how did you polish and learn your programming skills in Python without a computer science background okay so that's a very nice question which I always get asked now the thing is when I would do I need to share my presentation maybe see okay yeah so so I think am i audible yeah yeah so when I was in my undergraduate years I wanted to become a software engineer and so when we used to have the semester break I used to take classes for C and C++ and because I wanted to go there unfortunately the recession struck that time during a final year and the only companies that who actually kept their offers were the core companies so so I already had a job offer from a leading software company but the DEA their offer and then I joined in a code electrical department and so I used to like programming and I continued that and I had some background in C which was out of my hobby that was not taught it was just taught in the first semester so like I tell everybody programming is something that you can easily learn if you want to see for the programming if you see that way all that he required is a computer on the other hand if you would want to learn something about civil engineering and mechanical engineering you want to learn the practical aspects you would need to go into their labs which is very difficult to reproduce so it's just about your interest I think and I used to keep doing it as a hobby yeah true and everyone has to start from a hobby itself I guess I have the same background I also started as a hobby the question is like a lot of people have asked about the meetups that you talked about and where where can they find can they find the recording of the meetups is it possible to attend the meet up and talk that you that's tomorrow as well yeah so if you're not able to well if you go to the meetup page of Hyderabad the many machine learning you'll find that but if you're not able to make it we all YouTube channel and we all this post I'll record instead so we will also post links to all the meetups and comments to this YouTube video so you can go and take a look there and interesting questions so I think some people are not very familiar with women and machine learning and data science so one question is are there any chapters in UK and Europe yeah yeah there are in fact before India so it has global presence from us till UK living in Canada everywhere you have chapters so if you could go to that site of women and wealthiest Kawachi you will find listing of all the chapters over there you could start your own if you feel there is no one around you or you could join in existing one so she's asking she's a stay-at-home mom who wants to learn machine learning and she's studying on her own as much as possible and she used to work in ninety ten years ago and could you guide her like what would your opinion be on how to start with machine learning and data science after such a long period of gap and working with the family you know and she also wants to join your program ok so uh the first thing that I can do anybody who wants to come into this line that why do you actually want to come because I feel it's not something to you know discourage anybody but if you feel that you really want to become a data scientist because I think there are other have been usable so you could become programmers of a developer which are as exciting as this so first if you think that yes you liked it are you like of Tanjung into data and diving into data and then okay so once you've decided you want to become this then I always encourage people to polish their programming skills because contrary to what a lot of people do nowadays is you know they think that you know learning a little bit of fighter and this and this and that just jump into the algorithms I am little on the other side I always feel you should have a stroke a little strong programming background you should know what is going on and if you have to create functions and stuff like that it becomes easy so if you have a little so start working on your programming skills little bit and then you a little confident I think and then paddle early you could start with few basic courses that people have I think everybody take them in the entry line because yeah that's like the ABC inevitably starts with it and don't spend too much time only on Theory start with the practical portion also so programming his course you can take and then you can join a meetup near you and then just get in touch with us and then take it from there so don't think that you know it take ten years for it to start with just make small small goals and try achieving them first yeah I mean that's how you should do it right when you when you're starting something new one one of the questions is what are the what are some of the data science resources that you consume data ok so wonderful suppose all you say site is Twitter I get up and I actually scroll through now I only follow people on Twitter I don't follow politics at all because then you are going to get distracted I follow a lot of data scientists researchers and sometimes you know it the latest papers that come out you come to knows from from Twitter and I come to know what is happening which means it is happening what is what is the latest in this and that so that is a very nice thing again nowadays I rely maybe so there is a time that I used to also read medium articles and but now I mostly found one Kaggle and it's a lot of through discussions in kind of that I come to know and I scroll through the computation solution of like a be shaken put it out there and then from there I the worst things I do something that I could blame I searched yeah so my resources are the same it's it's also something that you have already answered but maybe you would like to provide it just what all resources did you use to start your data science career and did you did you do a master's degree in data science or a related branch do you want to do it do you think it's useful so I I did not do I just had my bachelor's only I would also want to do now because I think I am just I think now it's more about experimenting thing that my stage now I through masters also that two ways to look at it if you want to going to research in academia I think master's will definitely help you especially so I don't know if you're talking from India or outside but a good college maybe outside would definitely help you from going to research or academia but if you want to want a job I want to work as a data scientist I think more applied knowledge or would definitely matter and that I think comes from job experience when you work with people and maybe through Gaghan and stuff like that so I did not want to go into academia and research I did not do any masters and then I did not do masters because I was from a different fields that time I didn't know what I would end up this but ambition is China's past just maybe he could also provide some this is also very interesting question how do you manage between work and family and I am she's asking only about work and family I will also add cattle to it so when you're a mother you also have to decide you have to make sure that you're giving your kid especially like when he'd be small because you don't want to sit in front of the computer the whole day and then you just leave okay so for me uh one thing that has helped me is I had a little minimum distractions so I am NOT on any other social media except LinkedIn which also I manage this some time that I give to LinkedIn I do sit on it the whole day I wake up pretty early because it is that time I get a sleeping so that this is a lot of time to actually work without him coming and you know so by the end he gets the bunt down with my word and so I divide my day like in three parts morning when he's sleeping and then when he goes to school which nowadays he's not going because of the covet the show and one when he sleeps in the when he takes the afternoon not so that way I had divided my time and I don't know how it happens but I think method start doing multitasking and they become really efficient because they get they learn to work in minimal sleep so that is also here yeah another question I mean some questions are very much related to each other so I'm skipping some of the questions and if I've not if I've not asked your question then feel free to comment and I will ask parallel to find some time months of those questions what is it some some of these questions are really related to each other so I don't know how to formulate them so like do you have any kind of insights on women in women partnership participation in cattle yeah so actually my time notebook that I actually presented was on women participation in cattle which is which has talked to 16% from the last three years in this waiting for the latest survey that happens this year to see if anything gets changed so it has been 16% for the past three years and but the only good thing about that was is us in India the leading a lot of females who make the cause that 16% are from the US and India at the countries it's again very bad and in India a very important thing to be seen is a lot of students are active on cagin it's just we have to make sure that those students stay on Kerala and they just don't drop out from it so yes so if you want to know more what about that might we'll study was on that oh I could share a link and you could just go through that notebook yeah pearl it's a great notebook I've seen some of your notebooks really nice lots of things to learn so if you have not seen borrows notebooks then go to google.com slash viral Pandey right and click on notebooks and yet there are a lot of nice notes notebooks there's a lot of things to learn there are still questions coming but we are unfortunately out of time now I will just end this talk with one final question and you have to also guess who asked this question so what's your favorite joy okay so take a moment actually just tell people about this child time data show which is run by same Danny I think he's doing a wonderful job in getting a lot of females on this show and I think we should support such people who are actually thinking so much about diversity and I love all kinds of joy I am from hills and from the odd King behind I ease any part of my daily routine yeah thank you very much peril I'm not I'm not a child lover but I think the name would only want me to drink tea so maybe next time and thank you very much very nice very interesting talk and it was nice to have you thank you for taking the time out and thanks thanks again see you next time yeah thanks a lot for having me and if anybody was to get in touch Mike you can just get in touch with me on there social media handles or any meet up that I'm talking so you can you can get in touch with her on LinkedIn and I've shared some some links in the description box so if you want to get in touch with bar we'll do that and thank you very much for joining and see you guys next time