DevReal: Tomaz Bratanic Interview

Recording: DevReal: Tomaz Bratanic Interview

hello everybody I'm Alexi kber the founder and organizer ofi which is the oldest longest running biggest technical a up in B and the world and tonight we're at AWS CH Loft we have a massive M we have three talks and this is called knowledge graphs and agents and here we also have Thomas B who works with NE forj and uh writes amazing uh blog posts about graph frag and knowledge graphs and everything cool and fun and geni so welcome toas thank you for having me so uh tell us a little bit right uh like how did you come to to work in this area and uh how do you get inspiration to write such amazing blog post and teach a lot of people about knowledge gos so I've been working with grass for now like seven or eight years something like that and it's funny because like I only started developing when I was 25 so like no official background and I at 25 like I got my first job to do like connections mhm and uh my first boss was like very obsessed about new he was like new is the f fure it's like it's about Connections it's like not not data points it's about the connections between data points and like how do you like interpret the context like so like you on your own like I mean you might be interesting but it's very interesting that you're in San Francisco and you're talking with like me exactly and like we're doing meetups so it's not just you but like like your neighborhood and like what's going on around you right so that that was uh so 25 did you start programming at 25 yeah this is really cool how I mean like what what triggered you what's what's kind of switch Yeah I needed money okay that's the idea that because uh it's funny but I played poker before so then I was 25 and I was like kind of thinking like should do I want to do this when I'm 4 and like yeah maybe not like let's let's to get serious and like What's um like a good and my two brothers are both developers mhm just like maybe let today that and basically one of them got me like a job and that's how start so you didn't win a lot and bger yeah I mean it was like you know it's ups and downs cuz gambling is tough it's stressful uh but I mean it was good I it was enough money so that I didn't finish college this is this is an awesome story you know this is you're the second you know second guy in the community I meet play Pier so we had in the Skol committee we had a guy called Paul Phillips who won millions in poker and then started to work on Skol compiler as a fun project because kind of he always want to do compilers so but you know he was a developer before I think so this is great this is amazing so but like the fact that you played poker means you have like a grph of quantitative right yeah I me because it's also like poker also very like matte heavy like when you go beyond like just the amateur level it's a lot of about matte and like a lot about expectation it's like it's like it's funny um the most kind of important thing is that like U I feel like like if you you should just keep doing the right things it might not always end up U good right because there's some variant like you do a good thing like you do a good play and you might lose money but that's fine the important thing is like the process of like playing good and doing good and like over time the more good things you do the better it will be for you right and I feel like that's like a good um like mindset to have right so it's not like uh everyday results but like longterm uh process uh to get better every day it sounds like a great career advice for J to right it's like you may or not you know yeah I mean for at L it's like you never know what's going to happen right and like just build like a nice framework and like iterate daily weekly and improve and like eventually hopefully you get a million dollars from VC right was right a million just to start right so so okay so so you start uh uh coding and like what was your first program language yeah I actually cuz like I told you my boss was really obsessed about NJ so my programming Lang language basically wasn't really coding language but was Cipher which is like a query language for gra databases so the first thing I got into was like data modeling and importing and like doing um like basic data analysis and only after that I learned python wow this is really really cool and uh I mean it's kind of amazing right because so when I joined NJ I was like catching up because I knew about NJ for maybe 20 years like it's been around for a very long time right and so I was I wanted to catch up the first book I got was your book on gr data science and money right and it's like it's just fresh off the press in February right like like I bring it about show it to Philip R the C he's like tamash is a machine he makes all the content for us like and so like I knew and then now you write another book right yeah so how how do you how do you find time like I mean it's tough like books are tough because it's like wow it would be so good to write a book but then when you're actually writing it's like why did I do it right right that's right that's right but you did this right and so like the graph data science basically graph algorithms like for forj which is not gen but it this is all basic algorithm I think this is awesome right because nowadays I think a lot of people who come to AI they're not really necessarily going through CS education and so they not necessarily went through algorithms when you say graphs they think it's charts right most of the time so this is kind of funny to me right it's like you assume graphs are basic but no uh so so you like graphs right and so how do like how do you find this connection like when did you find that knowledge graphs can help with LMS yeah I mean I always knew that right it's like nothing new because how I see this like if you have nicely organiz nice data then that's awesome right and it's just like it's as simple as that like nice clean data M it's nicely organized and that's like the the powerful thing uh that you can then build around right and I always like like even before L for the thing I was also interested because before the lens we call this NLP so natural language processing so like the whole the the whole like uh building uh knowledge graphs from text or like um generating s statements or stuff like that that already existed even before the LMS it's just that everything was so domain specific and like you needed like a team of machine learning Engineers to build data sets and it's like and I was always looking for the sides like this is so awesome but like I don't know anything about pyto and like I don't want to build custom data set but it's it's still cool right and then like once llm came out it's like okay so now even though I'm not like a machine learning engineer I can do this on my own and it was like the the like a natural step to just okay let's let's do it wow now this is awesome uh and so I mean I can't help ask you right so your comes from Slovenia you come from Slovenia like right like graph people right some of the best graph people in Academia in kind of practice right so what makes it Slovenia like I think it's disproportional right like two great gra people from one small country so yeah but then if you say two it's like one in a million right because there's 2 million people in slovia but I mean like Slovenia has a lot of great developers I think it's like a nice developer Hub mhm and like a lot of people are technical so I would say son is kind of disproportional in Talent so it's not just me but uh and I I don't know why maybe because we so tiny right and like if you want to do anything you have to go beyond you go beyond beyond and like then it's uh you have to go out of like your bubble the comfort zone like anyway cuz like you walk five minutes and you're out of Slovenia right you guys are connected yeah I like history like you aerian Empire was like Multicultural right all the all the nations all the countries everything is in Slovenian and Italian right like super connected yeah everybody speaks three languages like it's just um and like uh so I feel like that that that's also helpful that and then like I don't know I mean for like I maybe bi but like people like money right so then you do like and I mean it's kind of nice when you find n like things that you like but they also make a lot like good money that's and like gen development is one of those things where there's a lot of potential yeah but you know what's so like I mean you're doing really valuable contributions because people really learn from your blog post right and so I wonder right so like um you know you you go and you teach people so what do you see like what works most effectively right because developers ask you and like it's it's all very complicated and you were doing it for a while so but like what's the kind of what's your like best advice to developers like what's kind of defin Works to teach developers how to do this like fil content yeah so for my approaches I don't like white paper so it's like yeah graphs are good it's like blah blah blah they we solve everything that you want to do it's like okay yeah but like show me code show me how to do it right it's like so every time I write a block post it always has like code with it right so okay this is cool and this is how you do it right so you need the Second Step this is how you do it and go test it on your data on your use cases m see if it works for you or not and like I'm not promising anything I'm just saying this is cool for me and it works for me and like do whatever you want with it so cuz like I feel like the the the Second Step you need to give people the ability to then produce it like because you know I don't know if you're a developer or not but like when I was a developer when I started something new like if I saw something cool and I could get it working in like 5 minutes even if it's completely like useless but like it worked it's like oh that's so awesome ab and now I can change this one thing and it still works and it's does a little bit what I want it to do and like wow that's awesome like because if you say like like if you do like white paper and say no gra are good but like you do it on your own right then it's like it's like too big of a leap to get started that so you need a little bit of like hand holding to get like the first steps and and then people get excited absolutely I mean I think in the Gen era I'm a developer right like I don't need to develop every day but I love programing and right I like I like play with stuff and I'm happy to be in devel right so like I think everybody should be doing in devel and gener I think everybody should be doing some coding and generally like I would not trust anybody who wrote a white paper and it's not coding right because they got secondhand information from other people right and so I think this is totally the right approach like you get to write what only if you call right yeah like for the management I I I I like I agree white papers needs to be there because there's also management right you need to sell and stuff like that you need to sell things to them but like I'm a developer so I'm not like that good audience for that I need code right and exactly reproducible code cuz otherwise it's just like you're selling me something and I'm not buying awesome well we're looking forward to your talk today so know go code and produce some reproducible code send it to us we're going to reproduce it and so the blocks are on towards data science right that's the med publication medium Yeah medium Yeah medium toas branish go find it Go implement it and we'll see you in the community thank you thank you