Lightning Talks

0 0

alright thanks my name is Peter Baumgartner i'm here today with a plea to you to please build dumb things i do not mean dumb in any derogatory sense a dumb thing to me is just something that you build for entertainment value or to learn something or for a creative outlet when you're trying to learn something and it's really simple the epitome of a dumb thing to me is when I'm talking to my mom on the phone and she asked me what I did last weekend and I explained to her like the Twitter bot that i built and her responses oh that's nice that's the epitome of a dumb thing so I'm going to share some dumb things that I've built hopefully to inspire you to just build dumb things and put them out in the world it's a great way to learn about new tools and just kind of explore the world of Python so the first dumb thing that I'm going to talk about is this Twitter bot that I built that basically generates random accolades for the City of Raleigh there was there was a time when this whole thing was coming about Raleigh was getting a bunch of accolades for a bunch of different magazines i thought well can i randomly generate them and have them make sense so like a lot of generative art and creative things a lot of it is total nonsense but there are some pretty comical ones if you if you pick and choose so some recent good ones these are all at least from the last week we've got Raleigh's in the top five for vetoing bandsaws so get those bandsaws out of here best cities for ill-conceived unbelting shower rooms I know almost all unbelting shower rooms are probably ill-conceived so I think that's a bad idea faking sadism I'll let you interpret that one and of course number two in humping soaps I don't know whether that's you know whether you want to hump soaps or it's soaps that are humping each other I'll leave that up to you but I learned a lot of things just doing this about the Twitter API I used NLT k and wordnet to basically try and figure out what forms of the words would fit into these these mad libs here and also this is just a Heroku a script uploaded to Heroku and then just use Heroku scheduler to have it run every hour so that's my first dumb thing my second dump thing is Santana bot Santana bot is really great so I got this raspberry pi for Christmas I was like what am I going to do with it I was starting to go through this some of the tutorials and the first thing you learn is how to connect a button and then like print out that you press the button so I was like okay well i'm just going to substitute printing out that i press the button with playing santana's 1981 hit winning so this dumb thing all you do is you press the button and all it does is plays the chorus from Santana's hit song winning that's all it does it's like the worst jukebox ever but it's great I learned a lot about electronics learn how to program with a raspberry pi and use a GPIO shell scripting and again the Twitter API this tweets whenever you press the button so i can share in the love of santana the pup potty tracker so my wife and I have two dogs and we have this problem where we forget the last time that we let out the dogs and they will remember for us by going to the bathroom in the apartment so I bought these amazon dash buttons on an amazon prime day they were only a dollar and you can hack those so that basically they still connect to your wireless but don't actually buy anything from amazon which is awesome so you basically say i push this button and i have a raspberry pi running that detects oh you push this button add this item to the database and that item is when the dog did that certain biological function this is a great motivator actually to get people to dog-sit I don't know what it is but people really like pushing the buttons it's fantastic and then I've got this web app you can go there and also keep us accountable for letting our dogs outside I learned a lot here to the Raspberry Pi as a server just have something continually running sniffing whether these buttons are trying to connect to the internet just use MongoDB to store all that there's a free mongodb host called M lab they got a free tier and then this is just a flask app with some basic bootstrap to get it all mobile friendly and whatnot and my final dumb thing is it's working this is great when you previously had something that wasn't working and it now works all it does is it plays this stupid 3 second video of Anakin Skywalker from Star Wars Episode one the Phantom Menace when he screams it's working when he makes his pod racer work so it's a nice thing when I'm feeling good that I finally like debugged something and I just go to this and I'm like all right Anakin thank you those are all my dumb things thanks for listening to me and please build your own dumb things and put them out in the world thank you so much no joke required um all right Julie is up now and then following her Peter parente up perfect great more technical difficulties I can see what's going on yeah you just all have to come over here stand behind them monitor yeah you could do in the laptop ram try unplugging it yeah I know there may be this uh system just doesn't like max what keep them entertained you want to tell some jokes James yeah you'll have to keep them PG ok g y'all have to give them G yeah ken are you up for it have you all heard the one about the broken pencil ok don't worry it's pointless ok cool so my name is Julie and I'm a PhD student in chemical engineering and since I have a captive audience I thought I'd tell you about my research so today I'd like to introduce you to how our academic lab is using basic data science skills in Python to improve bio processing so on screen is an antibody in our bodies produce antibodies in response to germs and a new type of therapy called immunotherapy is using antibodies to mimic how our body naturally fights disease and antibodies are really powerful because they specifically target diseased cells so you can you you can attach drugs to antibodies to deliver the drugs directly to the site of infection so this method is used to treat a lot of diseases like Alzheimer's arthritis and even cancer and we produce these synthetic antibodies in lab by reprogramming cells to make them unfortunately though we can't just program the cells to make antibodies they also make impurities and these impurities cause undesirable side effects and patients and they must be removed from our medicines and so that's where my research comes in so antibodies are large molecules now they're actually really small you can't see them with the naked eye but in the world of biology they're really big molecules and they're made up of tiny amino acid subunits in some of these impurities defer from the target antibody drug product wise little as one amino acid out of hundreds so removing these impurities is necessary and important but it's also a significant challenge so companies currently spend many months and many millions testing hundreds of filters and conditions to purify these new antibody drugs and they even use robots to all of these experiments and all these experiments have to be repeated for each new antibody drug because each antibody has unique properties in each production mixture has unique set of impurities so our labs goal is to harness all this data to help companies develop better faster and more economical purification processes for antibody drugs so my research will help guide filter selection I'm really interested in connecting the protein surface properties with how they perform in the filters and so I'm doing this using machine learning and other basic data science tools in Python and scikit-learn so as a model system our lab made a library of different antibody fragments that deferred by one or two amino acids so first we collected data on all of these variants by doing a lot of experiments like years worth of experiments and now we're using all of this data to develop kuser models or quantitative structure-activity relationship models and so we're connecting um we're using these models to uh to connect the protein surface properties to how they interact in the filters and okay I had some more about how like the specifics but apparently setup took longer than I didn't account for a setup time in uh in my speech but yeah so if you want to know more I guess come find me later to hear the second half thank you very much Julie it's very informative um alright so next up is Peter um and after that Eric you I know you're around yeah okay doc in my blind spot so okay yeah by design right obviously okay uh and yet another mac failure there goes my time it's ticking come on yeah time for another joke so why ask you got like jokes okay you'll see that almost all right so lightning talk five cool cool Jupiter things I think they're cool we'll see what you think in five minutes here we go ready okay number one how many people use docker like docker cool fudendo there is a set of Jupiter pre-configured Jupiter docker containers you download them you type docker run and you're up and running with Jupiter and you can get things with spark pre built in that's what this little animation showing the Allspark notebook has Scala our Python pre-configured and you got a Jupiter notebook server as soon as I go to the right URL in the video there's one thank you check it out it's on github too we did a tutorial talk on this the other day but if you have a nerd of it there is the dashboard extensions for Jupiter they let you arrange parts here notebook into like a little web application this video is just showing it happening right inside the Jupiter no Pope you can actually deploy these as a little web apps so you can see a little Scotch dashboard here for selecting a scotch that you might like give it another statute you've tried before so dragon moving around you could share these things pretty cool check it out next one you can actually with another extension out in Jupiter this is the Jupiter Colonel gateway you could actually annotate cells in your notebook with little restful verbs and endpoints and you can turn your notebook into a web server web api REST API this is handy if you don't want to build a UI like a dashboard you could actually have another person come along who's good to JavaScript and things and build your UI for you you just work on the the back end so Jupiter Colonel gateways the name of the project for that one how many people have tons and tons of notebooks and they're trying to find something to in the past and they can't because they got to click through all the untitled one through 1000 you know so there's another extension called the Jupiter content management extension you can actually search your own notebooks those full-text search over the inputs and mark down and things like that you can see it happening here it's a tiny video it looks better on a full screen obviously and you just click them and it opens and it's better than clicking through them all number five kind of combines again the content manager extensions once you search and find these notebooks min Reagan Kelly actually worked on this originally you can import them as Python module so you actually make note books that are reusable libraries of functions and you can search firm you can click insert into notebook see it injects is giant name which is a really long name of a notebook and I can start calling functions on it I think this video is a little long i donno if it's gonna get to the point but you could see it autocompletes and there we go and i got one bonus in case you didn't know i hope people realize this presentation is actually a notebook being hosted mb viewer and viewed on and be viewer Shinsei hosts hosts and github viewed on mb viewer as a slideshow so built into the notebook you don't have to install anything today you just go to view and you turn on the slide toolbar and you can actually turn yourselves into slides and another mins awesome extensions you can upload as a gist and then you go to mb viewer and we got this really cool inception thing going on we're watching a video of the slideshow that i was making before I don't even know how to explain it but there it is right so and that's it there's the links i can post it later i'll tweet it out thank you thank you Peter Jupiter looks better on a full screen um alright so up next is Eric and after Eric will have Christine you're right oh she and my blind spot still too she around oh there you are sorry oh okay you're behind me the whole time alright so I've got another bad joke to frustrate Eric with and I got this really excited you off the stage for i'm gonna forgot oh there there it is it works okay that's on your back just give me a terminal there you go go yeah oh well down saving me oh my god it's I've been mirrored I don't know how to do this now okay i'm gonna try and all right so uh download all right what is this terminal yeah it's super easy look at this all right ah ok bash all right so bash where where am I all right uh let's see if he has mini condo which I hope he does know it like it like seriously was here is not gonna go uh downloads mini Conda wasn't in downloads it's on downloads oh my god here we go alright so mini Conda I'm going to install this in in batch mode into temp mini Conda oh my god thank you guys I try to do this on a monitor that is literally 40 feet away because my screen if you can't tell is completely blank so okay so we are now installing and installing and more installing okay so export path equals temp mini Conda bin colon path so I don't nuke everything which Conda ok we are know that conduct great so Conda update dash dash all oh wait I actually want to you can regenerate this later move the condo RC so we know that is completely all right Conda info dash a just just to show you that i have no Conda RC yep no con Darcy ok so Conda update all dash see Conda for so we're going to overwrite the entire install of mini Conda from condo forge not have to probably shrink this down oh my god yeah it's so easy to read look at that why well no I buzz I'm not going to download it over conference Wi-Fi because it'll be the rest of this talk but basically you can see that everything is going to be upload everything is going to be overwritten by Conda Forge which is awesome so what else can we install from Conda forage tensorflow if I can smell it also not going to download it because it's very big but you can install tensorflow ah anything else that you guys care about that's it sits off the bottom of the screen oh my god Conda install Jesus all right no what is not okay so enough of that um I literally can't even get all right okay how active is this let's now okay let's see there it is crafts alright let's see if the spool up how much time do I have left like four seconds sweet alright so that is the commit graph for Conda forge you can see that this line is 200 and it's mostly due to this guy John Kirk I'm who's amazing yeah so super active it really took off around March of this year now how many packages are there again let me move this over so you can see the URL anaconda that org slash Conda forge there are a lot of packages 1159 packages which is actually more than anaconda then continuum will give you on their official channel by about a factor of two super family group of people you can install all kinds of great stuff from them and here we go Conda info Conda forge / Python no shoot shoot need the Anaconda client so that I can access that please go quickly yes that's shocking Conda Ford Python if you really want weird versions of Python you can get them on Conda for it including one point oh three point six alpha and to point out so with that I will leave you thank you thank you Eric you've never coded in Python one it is a trip okay Christine you're up next and after Christine bill you're after her so okay perfect I'm not going to use any slide sir they won't be technical problems I must statistics PhD student but I did my undergrad and masters in electrical engineering and to be honest I think it is sarcastic that I currently work in the statistics field for research and I hope to become a statistician after I graduate because when i was in middle school and high school I hated statistics so much starting in middle school I wanted to challenge myself to try to finish one year earlier and at grade eight the teachers told me that the test that I have to pass what to take the same midterms as the great night do and then if I score above their mean by more than one and a half standard deviations then it is considered a passing score and this is required for each subject and then on first try the exam was very easy and then I scored 95 over a hundred points but I didn't pass because the standard deviation was very high because like many students score high but some people didn't study at all and then the minimum required passing was 98 points fortunately the teachers gave me a second try and then I finally realized that they were helping me by making the questions very rare difficult so I thought I thought that difficult questions were very difficult and felt frustrated and then I scored only 19 but it was a passing score and the teachers even told me that you only need a 70 something to pass so I went off and finished one year earlier in middle school but the problem was I started to hate statistics because it was I have the same level of same level of ability but it could be changed so so much and then in high school I continue to hate statistics because we weren't allowed a calculator in math and then the teachers tried to challenge us by giving very difficult numbers and I thought that was very complicated and I thought and then I didn't like it so I thought statistics was used to torture people and make people feel bad then I went to college and I majored in electrical engineering I left I like applied math but I decided not to do the math major because I knew if I work if I do a math major I'll have to take more statistics than I was interested in but in the electrical engineering I missed one point what was that I mean miss one thing is that in the sophomore year we had to take a required course probability of status and statistics Hivers things started to change because I realized that the standard deviation is just one line of code no more cargo no more complicated calculated calculations we can use a calculator and we're also I don't also learned to solve real-life problems such as signal processing okay it is transmitted during a noisy channel and then we can see like it 0 but when we observe that as one we can calculate the false positive rate and the false negative rate so things became more and more interesting then I went on to take more statistics and applied math courses and in my Master's I even worked on thus how to compare distributions on Facebook beta and by the time I was applying for PhD programs I had a big mental debate on whether i should do a statistics or a applied math or a scientific computation PhD and then i finally decided on statistics because of two reasons the first reason is that it doesn't require GRE subject tests and the second one is that if i choose statistics there are a lot of options that i can work on for interdisciplinary projects and four years later I'm glad that I made the right choice because it opens up for a lot of new process possibilities and applications the lesson that I learned is that be open to new fields and don't be don't own don't be limited to the past bad experiences because you never know what the new field can bring you opportunities thank you thank you christine all right bill I think you're up and then that's I I think we're out of time though right isn't it keep going all right if you guys are okay we have more time in for that so then the next person ah is Charlotte so cool all right go ahead take it away awesome good afternoon i'm bill newman and i'm very new to the data science space but I'm not new to IT I've been in the IT space professionally for just shy of 30 years and in that time almost almost 20 of those years as some flavor of database administrator or data database developer so I'm speaking historically but it may not necessarily apply because I'm new to the Python Python world but history doesn't repeat but it rhymes historically developers have used the database or data storage as just a place to persist their data and they have historically done it very sloppily and it's always been a hard-fought battle sometimes almost to blows but but we've won out and developers have a sense of data discipline data management but now with the data science moving away from the IT space and being more closely married to business managers they are falling away from the IT oversight and I think it's a whole new a whole new genre of developers and it concerns me that I'm starting to see some of those same problems with big data because you can put your data and brute brute force it back you really don't have to worry too much about how it is or where it is because the hardware will see you through you'll get it back so I it could don't lose out don't forget your data discipline and data management skills keep them always in mind additionally I come from a military background and let's speak as far as information and the tools that you build it's true in intelligence work that the data you grab is unclassified until you develop an interest in it and compile it and now it's classified let's go with the the predictive models that you build you spend all your time tuning them getting them just right it's good and now in the enterprise you have built a decision maker which is very valuable but historically developers don't come from a security mindset they're less concerned about securing those items securing the tools they've built and it's always been again a hard fought battle to take the to teach them to secure their products to keep them contained in a secure environment and to worry about security access rights but again the to the environments that you have are now your own not so much in many places with IT oversight so you build a decision maker that's very good that your organizations rely upon but now put some mind to securing those things and protecting them because if you lose them if somebody grabs it now they know how your organization is going to make its decisions because you handed it to them so again i may not i'm a i'm new to the space so this may not be as much a concern because i haven't had a chance to see how things are developing but i charge all of us to be mindful stewards of the things we build and the things we have thank you thank you so much all right so next up is Charlotte uh also slide lists oh thank you so much now is worth uh yeah great lady talk oh thanks come back next year all right so I just wanted to get up and talk for a little bit about pie ladies and community involvement so it was at first I want to make a correction from the closing slide I did not actually get a job at the conference it is true that I am starting on Monday at cactus and I'm very excited about that but that did not get the job here per se anyway um so that being said another life changing event has happened for me in the past couple of weeks is that I was asked to become the primary organizer of Pi ladies anybody who is a local pilot ease anybody who's familiar with the local pilot ease may be aware that it's been a very quiet group lately and that is largely due to I had two primary organizers one of whom just had a baby and the other one has just been over committed and so it was decided that it was appropriate to pass the torch and so I am going to take the lead and we are working on revitalizing the group so if you've been aware of Pi ladies and kind of dismissed it because there wasn't much going on that is about to change so anyone who is interested in being a part of pi ladies I encourage you to go to the meetup group and there's going to be a lot more happening I'm really excited about the directions we can take it that being said I also wanted to mention the tri Python group and it's been mentioned a couple times already during the conference pi ladies in this area works closely with try Python and we hope for that to continue and I also wanted to mention it for those of you who are not local or even those of you who are local look in your communities and see what groups are out there because it makes a huge difference in the strength of the community as just having a conversation with somebody earlier today about how one of the things that strikes me about the Python community it is very much a community you have been to different kinds of user groups and all this sort of thing and I'm always struck by how much the Python communities are that community it's not just a place to go and share knowledge that certainly happens but there is that camaraderie and people working together and supporting each other and that is so important to nurture that so I very much encourage everybody to go out and find whatever your local community groups are get involved show up even if you don't have much time you can still show up for a meeting now and then and you never know who you might make a difference in the life of you know to throw back to Carol's keynote speech speech you never know who you are going to affect with even just a short conversation so I encourage everybody to get involved in your local community even if you think you've got nothing to contribute I would tell you if you'd asked me a month ago I would have said I don't know i could probably contribute something but yeah I don't have that much I don't have too much background you know what here I'm standing here becoming the primary organizer of pi ladies because people liked my ideas and wanted to see my enthusiasm go into that endeavor you never know what you're going to effect thank you so much Charlotte all right now we have a JC and there there may be magic apparent yeah where there is magic okay um and then following JC we have Jason so Jason okay yes around James J no we're skipping James as Jason ok Jason's on around so let's give him it don't the kids are gone to so uh uh uh and then humans at PI data I think yep which is Lucia so hi everyone just to start I have like no now a free personal battery great but I'm gonna clean everything just to make sure it's tough on a clean plate and what I'm gonna do I going to deal the wheel and just show you how you build a wheel I'm gonna do a I'm going to time it we're going to take a minute and 30 second a wheel and I can't just going to a parallel below that with a 42 42 why not okay because you need to go in the virtual environment site done on C Python on wheel yes and ok and here we are building a C Python on wheel it's downloading Python it's configuring doing the introspection and and then you're going to build the thing and and yeah and then and then they're going to show you in the meantime today I i put my bug socks and i'm pretty happy about it and yeah yeah and like what oh it's going very fast yeah I don't have 42 cold butter but it's pretty good like set up a nice um and it's going I've 64 gig of memory on that mansion by the way and oh yeah yeah it too much but I should have taken off cough to be able to use all the memory and the drive is pretty cool too envian me and I've like nan Jie Jie Liang gigabytes in reading and yes it's cool i mean you can do is done things like Peter was talking about doing dumb stuff great i'm getting the wheel now you see C Python will awesome let's check if I got the wheel if I got the wheel yeah and and I of the let's clear everything to make sure you sit and I got the wheel like here but now let's install the wheel and I'm gonna go in test and I'm gonna clear everything and I'm going to make sure that I don't have anything installed yet northern OC Python will are going to install it pip install and I just give the path to the wheel gonna make things a bit bigger make sure you see things ok installing oh it's installed great I'm gonna go go back to something bit smaller because it is more things can happen on the screen and then I start Python ok its version 2 7 10 no big deal oh I mean you can see it 27 10 like here and now I go back small it's fine out your back because I have time you know and then I do import see Titan as a SJC why not and then and then I gonna do JC dot take over and then I wanna happen well I mean we build that like just before the talked and then it's like that's an ascii art i'm not sure it is but i'm adam it's cool you need to add do I think before I look you look something it's like cool things to do but but now what do we have what is that thing yeah it's pretty useless but now let's just do import sis and let's do sis that version and what do we have here we have Python I'll you come let's go let's go let's go big now yeah and we have its Python a 3.5 phone too and it's very real 3.5 photo is not just a string and printing on the screen too yeah and it means yeah you can build a we love of C Python cipa there's no we love C Python we could put it on pip could build it with with many linux and I've will for any version of a fight on that's it and yeah and now to show you that it's it's oh yeah i just did control the exited the first one and now i can take over again and i can leave again you know that's not it and okay now let me show you how does that things work clear and i have a set up that py pretty standard stuff as can you see things the red now you can see it can you yeah no yeah no okay let's change hey let's i liked it look at that it's better no hahaha no anyway I gonna talk and you're gonna just trust me and that's gonna be alright yep you have the version 3.5 point too we just specify few option the package description metadata like a regular wheel and that's pretty much and in the folder what do we have oh if I can type I guess I and in the folder I just have a the source and my set of the py and and that's it voila thank you so much JC hello everybody if you still do not know me that means I've done a great job at letting all the speakers and presenters do what they were supposed to do which is give you lots of information inspire you continue to help you pursue your career and your dreams but I also wanted to give you a little background off what was actually going on behind everything here I'm starting off with a few pictures and I'm ending with my most favorite part the numbers the heart numbers first of all it all began with two people at a bar it was actually not a joke it was two people at a bar after Evita and Chris came to me it was like you know what I think we should get the Python conference here and I was like I was naive enough to say sure count me in after that we had our very first board meeting with a very big board we wrote on itself ideas on and you can already see a few of the face as you've seen across the last three days guiding you along the way helping with the setup making sure everything runs smoothly afterwards we kept the momentum going we had more and more meetings and then what happened that one of our report meetings the man bailed out you know what there was not a problem for us we kept going I guess what the men joined us again and we needed a bigger table then one day we got very excited because we had our webpage life so we were like stuff is getting real right now we need to hold it now because we promised the world because we announced it on the internet and when you have something in the internet it must be true this is how by date I Carolina started last but not least we did a ton of setup work and I'm so grateful for everybody having join us and the days before to make sure everything is set up and we finished two hours earlier than planned on Tuesday evening that means we got to eat and actually go home and sleep before the conference starts this is why you saw so many friendly faces starting Wednesday morning because everybody did have a good night of sleep which is apparently unusual for a conference organised here are some tips and here are some statistics on making sure this conference is reproducible first of all send emails early have conversations early it turns out the the very first email in my inbox with the word PI data was from march to was from april 2015 apparently were quite low in organizing a conference between just me and chris we had about two hundred forty three different email threads going back and forth over the last year or year and a half longest email chain was about 57 emails going back and forth the toughest subject we did have to address that nearly made us cancel PI data Carolinas to be honest or chp to and I'm glad we did not and I'm very excited that we were able to focus on diversity making sure we have a safe environment for everybody who joined us we also have an interesting statistics I was doing a little bit of sentiment analysis on the emails received that it turns out we do have a PO lightest email sender which is very Cleary a wonderful and amazing AV support the latest email send was around three a.m. at night now my colleagues please cover your ears my work colleagues should not hear that but the earliest email was 523 am so someone is zoomed it was actually the latest email upset in addition to all those stats we had so many volunteers and we could not have done it without the countless volunteer hours that everybody has put in so to all the volunteers here I give you tons and tons of gratitude and please come to me afterwards for a token of appreciation and some nerdy things to feed your Python knowledge also that one's for you Chris here is one thing you can definitely expect from us tonight which is taking a good night's rest so we will not send any emails until like three days from now because we're sleeping thank you all for coming and I hope to see you North Carolina again next year