LlamaIndex Webinar: How to Win a LLM Hackathon

Key Takeaways

um yeah I guess we can we can get started uh maybe with some introductions uh do you want to kick it off Raul well very nice to meet everybody and great to be here and thanks he for inviting me uh my name is Rahul I I am a founder of a company called AI hero um I you and I go a few years back where we actually met in a hackathon so I think it was 2019 you remind me if I'm uh mistaken it was the first wave of hackathons with the chat uh chatbot era and we built Mr Miyagi along with Yulia and a few others Mr Miyagi helped people uh just navigate life life challenges as as as if you remember from the movie um and so we we got to the finalists round and that's where I kind of uh got hooked on to hackathons earlier this year uh Alex and I turned out to be in the same hackathon the Emir there was an emergency uh hackathon in San Francisco when Chad uh when I think open AI released its apis um and Alex and I were finalists in that Alex obviously won I'll let him speak about it uh Alexa and his team won uh amazing team work there and then um I took part in multiple other hackathons won a couple of them myself um and I I really wanted to um join the panel and speak on how exciting an opportunity it is for anybody participating in the hackathon to learn and not just uh participate uh passively uh one of the ideas that I got from the hackathon myself I converted into a product or a tool that I I wanted to show not because I'm pitching the tools so feel free to ignore the tool part of it but I definitely want to show you guys how I how cool I uh streamlit has been as I use it um so this is this is a tool called prompt craft and it in this case we are going to create a completion llm completion um for English to Japanese so the moment I change my prompt template um you know the variant gets automatically saved you can see all your variants on um online you can use a variant that you previously had of your prompt in the future and then this also allows you to plug in any of these variables that you've added in and tried to complete it what I love about streamlit here is um there is got some aha moment for me was you could uh use streamlit in a progressive app manner so if you change stuff on the left or on the top of the page everything bottom changes here you can see I added something we are adding some tests um the tests run automatically as well and you can see them in the UI and compare tests and make the right call about your prompt and iterate on the prompt so again not a tool plug just showing you how awesome it was for me to use streamlit and you know some ideas like this can come out of a hackathon that's it cool and around how many have how many have you been to now oh I have like it's gotten slow recently because I have a fatigue of hackathons uh the most recent one I attended was in I think last month and almost one every month a couple of every month yeah and you all you've also organized you've been a finalists many times you've done a bunch of presentations yeah yeah that's right so I have uh organized the hackathon for the envelops community I I am a local organizer uh here in San Francisco we have a few meetups coming in so if you guys are interested in learning more about llm Ops and uh ml Ops um please join us cool all right next we have Alex hello hello uh super excited to be here so I love talking about hack funds um so just kind of a bit about my background uh I've done hackathon since uh 2016. I absolutely love doing them it's kind of like an arena where you get to take crazy ideas and then smash somewhere between 8 and 48 hours in Savannah and just build a cool product and just validate something fast and just play around with cool technology uh and over my career I've won 16 of them and it's not as hard as it sounds uh in fact I'll kind of give some encouragement to those in the room who come from like a less technical backgrounds uh during my undergrad I actually uh did not study computer science I studied philosophy uh that said I know how to program being technical is definitely goes a very long way forward but uh just you don't have to have like this extreme background in computer science and programming to learn how to win a project uh and the number way one way you win these things is uh you replace the word project with people and ultimately at the end of the hackathon you're trying to build something for people and the way you win these things is you just talk with people figure out what a real problem you can be solving this and one of the wonderful things about these things uh is you have a lot of sponsors and you have a lot of people in the room who are basically paid to facilitate and Mentor so instead of spending all of your time working and focusing on this building product and heads down in code I highly suggest just take maybe three or four hours and just chat with people figure out like you know maybe you have an idea or maybe you don't have an idea and talk with a mentor to talk with the people who know more than you do often people who are from very strange Industries and kind of get an idea of the problem space and then with that you can formulate a hypothesis into or am I solving something that's real that's going to help people not not a cool product am I helping people and that more often than not leads to victories so uh because you're solving your real problem and you're able to show that you've solved it everyone's gonna want to vote for you because you went out of your way to do something that's awesome um yeah and just kind of some cool things I'll tell you like why you should do hackathons in general um I've just won a lot of very cool prizes uh I'd probably say that number one was I got a job offer out of winning Hat Club uh and maybe title for number one was I want a hackathon hosted by Mazda Motors you know the car company uh and they flew us out to Japan for half a month as a prize so uh you never know what's gonna happen with these things um sometimes we get International trips sometimes you get jobs sometimes you get very cool uh credits things like this uh and uh there's like a lot of cooperatives that arise with these things so highly encourage it talk with people uh and it's a really great career booster to be cool people along the way and I think you all are gonna also make some great friends along the way yeah and just quick interjection there I also got my job at a hackathon so it's more common than you think um great great advice Alex so um uh next we'll have Caroline she's uh organizing a hackathon right now so if you're not doing something this weekend she'll tell you all about it awesome I'm really excited to be here really nice to meet you all virtually my name is Caroline I work on developer experience for streamlit which if you haven't heard of streamlit it is an open source python library that allows you to create web apps directly from Python scripts so really quickly really easily a lot easier to use than something like flask or using just HTML CSS JavaScript Frameworks stuff like that um so as a result really great thing to have in your back pocket for a hackathon when time is of the essence and you want to get from your idea to a prototype very quickly we are running a an llm Focus hackathon right now it is ending on September 19th so you still have a little while to build something awesome and submit it we're partnering with llama index and a few other companies uh for this event there are some awesome prizes that you can win I'm gonna drop the link to the page in the chat definitely check it out um take some time this weekend build an app like I said streamlined apps are very easy to build to get from your first line of code to a real app that you can play around with it's super fast um so yeah really excited to be here I have run a couple of hackathons for streamlit and I also work with our community with students who are interested in running um hackathons that have some type of Streamlight component in their community at their schools at their colleges so if you're a student who's looking to do something similar definitely reach out to me send me a DM on LinkedIn and happy to chat thank you oh yeah um and we're really excited to be co-hosting the streamlit hackathon um being one of the sponsors um if you build something cool with llama index you're in the running for a set of new airpods so um yeah definitely look forward to seeing all the cool projects that come out um so my background like I said I actually got my job at a hackathon I met Jerry at the you know I met him at the uh the same one that you're mentioning Ronald the uh the emergency chat ubt uh API hackathon um that that Alex won uh I didn't win that but I guess I got something out of it also um so uh I guess the first sort of question um for you is like Alex you mentioned like picking an idea right I think people always think about hey you know like what what do you what do we actually want to build there's a limited amount of time um whether it's a one day or uh you know even here it's like maybe you have a couple weekends really to build something like what's your decision process when you pick an idea maybe Alex you can go first so there's kind of two strategies um so uh the The Motto that they use to Boy Scouts is be prepared so uh number one thing you can do is it's kind of like I would say unethical to pre-code or your hackathons but it's not unethical to gather knowledge so definitely talk with people about your ideas beforehand maybe a week before um and just kind of float out like would this be a good idea would this be a bad idea and visualizing what that looked like uh one exercise I do very frequently is when I go on walks like I love to do evening walks is I just kind of think about ideas and if I can come up with one good idea or stupid idea that's like oh that'd be fun if it existed I write it down to a Google sheet and I have this Google sheet about 350 ideas a lot of them are really stupid uh but it just kind of like gets you in this mindset of like oh you know this this would be fun to filtering a hackathon right like I think I could see this happening uh in lieu of that if you don't have a big catalog of 300 ideas uh what you can do is just again speak with mentors so I did this one hackathon it was the New York Maritime innovation hackathon Maritime means like boats like cargo ships I I've never even been on a boat before uh and I was like what am I gonna do uh so I just ended up spending like two hours talking to these ship captains and like Logistics coordinators just like about the problems and like the shipping industry and just by doing that you very quickly realize like they live in a very different world than us in technology technology everything's efficient everybody's optimizing things shipping it's paper right it's physical paper and you only get that sort of knowledge by actually speaking like abandon your predispositions and just like validate ideas by talking to people who are the experts in the field they're going to give you much more information than you know anything you can just come up with on your own so just highly talk you know highly recommend you know hey get the creative juices flowing idea way beforehand and be talk with people uh they're gonna be your best source of inspiration that makes sense 300 ideas that's a big Advantage coming in so so Raul what do you think yeah yeah I want to follow that I think Alex mentioned an interesting thing which is you know you might some of these ideas might be stupid to be honest I I sometimes feel like in life I when I'm working I'm not able to try these stupid ideas right but when I do try the stupid ideas I can learn something completely new right I can gather discuss it so personally even though you know uh while the day on the day of the hackathon maybe two hours in we will I will start like honing in on hey how do we get uh winning but eventually what I really want to take away from a hackathon is well what am I learning right and um that gives me the joy of participating in hackathon you go and talk to people around you see what other people are doing you meet some really inspiring people um you you can learn from what they're building uh not to copy them obviously but like get inspired by that uh feel free to get your hands dirty on some new technology you haven't read a great example was um on in this in in a hackathon a couple of months ago on which Alex and I were on the team um we had another uh uh member Sasha uh I built I spent like two hours building something with a flask and uh an HTML I'm sorry a rest endpoint and she implemented uh alternative to it with websockets based on her previous experience at uh hackathons and that kind of I blew my mind like hey it's this simple to do web sockets for uh for a one-to-one kind of an app and we we completely had lost so that's you learn something new by just observing and learning with people and I think to me that that is uh the if you don't win the prize which you might only have a handful of them this is something you can definitely win in and what Raul I remember when we did ours we pivoted in the middle right and then I remember even at the emergency one you were actually going around searching for ideas for a little bit yeah so let me give you a background on one I think when we pivoted during the 2019 hackathon uh it was because it was an overnight hackathon right or we were working all night into in in a in Menlo Park College um and you know there were a couple of things we could take this to and then based on what we heard um based on what the team is feeling you know at the end of the day you want your team to be happy right you don't want to be like hey we're just going to win but somebody's unhappy on the team I don't think that matters we we I think democratically voted for the right thing to do um and then during the emergency hackathon where I I saw you after three years um the I had two ideas that I came in with which I wanted to try and I think my canvassing of like hey who's doing what was partly for finding a teammate and partly to just be like hey which idea do you think will be cool and the moment you get the public opinion of which idea would be cool to to see um you could go ahead and and and and you know see that okay there will be some Traction in in at least if there's a voting round you'll you'll probably get some good votes from yourself attendees makes sense and Caroline hey having organized a few of these already like do you do you see any patterns in terms of which ideas tend to work better than others yeah I think the number one thing that I'm looking for as a judge is uniqueness so new ideas that we haven't been seeing in all of these projects um you know a good example with the llm hackathon is a really common streamlined app around llms is just using the GPT 3.5 API and kind of making a chat GPT clone which is great but it's not really something that it's not something new it's something we've seen a lot um so really when we're looking at submissions we're looking for something that is kind of new a new idea at least um something that really makes the judges say oh like I don't even thought of really doing that in a streamlined app um and then I think kind of to Echo what Alex was talking about um really having doing some kind of research and having a story that your project is telling of what is this use case who the person in real life that's that's using this app or that's that could be using the the finished version of this project down the line so instead of kind of more of like a toy example um really making it something where someone can look at it and say oh I can totally understand how someone in real life would use this yeah the uniqueness is a very big factor um and the sort of like feeling that this is actually usable now given as a hackathon it's not gonna be perfect okay don't worry if you know if you have to put it on Rails right and you know you only give it certain inputs that happens all the time but um people want to see the vision something different and it's something useful um so I think well you you talked a little bit about this so maybe you can talk about like finding a team do you generally go into these with teams or or are you looking to find a team on day of or yeah and it's it's evolved after participating in multiple hackathons the first couple of hackathons you go in fresh you don't know what you're working on um and I I'm like I think with the most recent hackathons um Sasha and I we met at one of our um groups that we are part of we've met at a happy hour and we we got talking and we were like hey there's a hackathon coming up do we want to take part right and that that works to be honest better than me I I love this um group of friends who participates in every hackathon and they are just um they're just there to participate every time I've seen them in every hackathon they're having fun they're they're like learning something new and they're building up what what their resume is so these are undergrads who are who are participating I think that is something that is enviable to me which is you know uh working with a good group of people that you've collect like met and you you like you love working with and build something new uh that being said you know you shouldn't start by thinking like hey who's a great startup co-founder and say okay I'm gonna take part in the hackathon with that person I I think uh starting a company is a very different Endeavor than doing a hackathon do a hackathon with people you'd you'd learn something with and enjoy working with and you know it's going to be very intense eight hours a very intense week just make sure you're having fun you know and Alex in your 16 victories do you do you generally partner off with the same people or uh so for a bunch of them I have a crew so I've met the direct thoughts I'm like hey let's do another one uh that's always fine but I've done the whole gamut of things I've gone solo um I'd highly recommend against going solo uh because it's a slog it's very hard um I wouldn't recommend it it's just an enormous amount of work that you have to put in the pressure to put on yourself um in terms of just what you want to work with I just recommend finding people who you Vibe with ultimately it's going to be a test not so much of your skills but as much as your patience as you are in the late hours of the night two in the morning you're gonna be doing something stupid like data labeling and you want to make sure that the people around you are it's gonna go insane uh or maybe you will go insane but you want the people around you to help you not go insane so it's really just kind of a Vibe check um I have kind of a follow-up question on that how would you approach it if you were to go back in time to your very first hackathon you don't have a crew of people I imagine that can be really intimidating for a lot of people who are brand new to hackathons how do I find someone who I'm going to mesh well with we're gonna work well together they're not gonna leave me high and dry and this isn't going to be like a sixth grade group project where I end up doing all the work and they take half the credit how do you kind of navigate that in your first hackathon so I think like also for a lot of first hackathons you might not be a skilled or as experienced as a lot of people and that also presents a challenge like okay I'm not good enough to be on this team uh I'd probably say like that's when they present yourself in the way of like looking for somebody to kind of guide you to help with things um so I'll say this to like the non-technical angle if you're like non-technical um I wouldn't lead with that you just say like I'm trying to learn so like don't say you're not technical say you're I'm not yet skilled and I'm trying to progress to a point and say like I'm trying to get this skill during the hackathon I'm happy to do whatever it takes to get the project where it needs to be I think like somebody who's a team leader is going to understand that that you're going to be high velocity and uh just be able to uh pick up the steam from there um so yeah just basically if you're if you're kind of nervous about things just find somebody who has a vague idea of what they're doing and just say like I might not be the perfect person for your team but I'm going to do whatever it takes to do the small bits you know even something like um making a designer picking a color scheme or um I wouldn't recommend just doing a deck all alone that's maybe 25 25 of it but like uh definitely just like say here's a very specific thing I'm going to try to do extremely well at and that's going to go a very long way do you guys have a name for your Alex do you have a name for your team your crew calls themselves um yeah not really um well one of the hackathons we did was uh hosted by the next web which is like one of the largest Anthology conferences in Europe and we won that and they flew us out to Amsterdam for uh two weeks so if you go to that so we just kind of name our WhatsApp group the next web uh it has no meaning to it though how many people are in the crew uh it's about four of us uh and actually I'll post a little link uh because this thing was hosted by a hotel company called Trivago and they did like a whole blog post about our crew for winning this thing uh that's another thing by the way like you get really good PR so if you ever want an article about yourself on the Internet or you want some like you know internet cloud or you know take Internet points uh definitely you're gonna get some articles about you if you built something cool yeah um with the with the with Carolyn's question about how do you go and find people if you're the first time um at a hackathon I'm I'm curious ye when we met during the hackathon a few years ago you and I didn't know each other right and uh nobody on the team knew each other what what do you think happened there which which got us not just for that particular um hackathon but then you know we've stayed in touch after and we like I I Julie and I are really good friends as well what do you think works well with like what what do you think is the glue uh over there yeah so so so amazing that was actually my first one ever I've never like you know even though I went to college like it wasn't as big of a thing I'm a little dating myself it wasn't as big of a thing uh back then um so um it was my first one ever and I think I just got really lucky to be honest um but I think the thing that worked well um and the thing that I I think um you know the other one I won so so I won the one with you Raul and then I won this other one at uh Shack 15. the thing that I think worked really well in both cases is the sort of like complimentary skills right so so to Alex's Point actually you know if you're non-technical you can be a really huge asset to um to a lot of these uh hackathon teams right um I got so tired of seeing just the same chatbot interface every time that I ended up hosting and thanks for coming Raul um Rahul was a mentor there hosting this anything but chatbots um hackathon just because it was you know like I think the design the presentation um the sort of once again like helping people actually Envision that this could be a real product that's really important and um so yeah the the one with you I mean I think we had a couple technical folks and a couple of non-technical folks and that ended up working really well Rahul is a great presenter um just locked into it really like like I literally just locked into it but like I think the presentation probably made a very big difference for us um and um yeah it's it was you know it's sort of like having like different skills coming together rather than say hey you know what I I can program python you can program python you you know like having three people who are just good python programmers doesn't necessarily make the best team um and actually you know that does bring up the next point which is like you talked about this in terms of presentation Alex like what what would you say in terms of like the split is between how much you should focus on the presentation and and the sort of the UI versus how much you should focus on the technical stuff yeah so definitely UI makes the magic happen uh the number one thing that will make people's eyes pass over is showing command line so well okay so if you do have to show command line I highly encourage you if you zoom your screen and people in the back will not understand a word on that screen um having a good ux and you are having a great UI is really kind of a good way to visualize what's going on that's actually why I think streamlit is a fantastic tool for this uh because you have horrible front-end devs like me who do not know how to make nice looking uis uh Streamlight just doesn't pretty much instantly so if you're trying to put a date on a bid on a web page definitely use that um but secondarily like uh there's a few resources out there I'm going to post the link in the chat to this app called slides Carnival having like a good slideshow just to like 30 seconds to one minute explain the idea explain the problem very concisely it was a very very long way um in terms of time commitment I would say like maybe spend like the last four hours like working on your slide deck and just to make sure that it's a very concise message to what you're trying to sell um one of the biggest challenges I think I've seen a lot of hackathons you have very very technical people and they're building just insanely smart and Brilliant products but they have a very hard time communicating with the building uh ultimately again like replace products with people um there are people the judges in the audience they're your customers and you have to really relate to them and explain the pain in a way that is going to be obvious and uh they're gonna understand immediately and be able to summarize and you know a sentence like what what does the problem they solve this is how they did it um that that really goes a far away so I'd highly recommend just like having a good UI that explains the flow of what your app is doing having a slideshow is also very helpful for explaining the problem Set uh and that's usually most of it also rehearse right uh if you're a non-technical but you know you're a great speaker spend some time writing down the script and rehearsing the script and making sure you're selling the message so definitely highly recommend that a last thing though I will say like kind of a hacker code of ethics is stay away from slideware which is just like a slideshow with screenshots or no no code whatsoever like I think people really do appreciate raw product even if it's buggy um it goes a long way yeah and do to follow up on that I think the um one one where like one hackathon where I felt really um like bad about what I built out was I couldn't show the demo at the end and you know it worked like maybe 30 minutes before the demo didn't work at the time of the demo when I was presenting to everybody and I think that is fine I don't think I lost because of that right but I lost because of was that this this demo didn't didn't spark the Curiosity didn't didn't inspire people or get people the hook right like I think the hook will get people like paying attention but at the same time you've got to inspire them there's so many um you for example if you're a project or that you're working on the hackathon is let's say something for social good right great that's an inspiring message but it's also about how you inspire people with your slide deck so first getting shortlisted with your technology with your um you know skill set and what you've built um you know the judges uh would would would shortlist you bet on based on that but when you're presenting I think the presented like a good presentation steals the cake right like that that is definitely uh well taken I'm curious from Callen side like what what do you think are from a judge standpoint or for this hackathon maybe what do you think are some of the uh things that people look for yeah totally I think that one thing that people don't necessarily think about especially in a hackathon where you are submitting an app or you're submitting some sort of somewhat finished project that uh the the judge is going to directly interact with make it as easy as possible for the judges to understand what they should be doing with that app so the flow of oh I should be going here I should then be clicking this I should for example if you have let's say that your app takes a file that they have to put in a file uploader give us a sample file that's you know embedded in the app we can download it from the app and then we can upload that up there and we don't have to like create our own you know you don't want to say oh you need to go create a CSV and add these things and then you need to go upload it and then you can kind of demo the app there's too many like potential points of failure for the judges to not do it in the exact way that you were thinking and then it's not the process that you were hoping for them to see demo there um so really like make it super simple for people to interact with your app understand like the flow they should be following ideally that's clear just from like the design but you know if it's not clear from the design really outline it for them to ensure that the judges are getting the experience that you had envisioned and I think just minimizing those those points of failure where someone might be like I don't know what I'm supposed to be uploading here I don't know what I'm supposed to be doing here that's great advice yeah keep it simple for people to try um shubankar am I saying that right uh yeah uh you're right so I asked quick question uh during the team formation even in any hackathon which which I have been to um sometimes I have seen like you know you approach people with idea and they might not end up like teaming with you but you know they can take that idea anything with other people or you know maybe themselves go ahead with that particular idea so when you pitch your idea or during an information event when you want to like bring in people to collaborate and work with you so how would you actually you know um motivate them to join your team or in other words like not spill your idea to anyone else that's really important Alex you have any tips though uh yeah it's uh can't beat him join them uh so at the emergency chat hackathon uh my original idea was uh basically creating an AI Furby these little like bird toys I don't know if anybody remembers them from the 90s uh and uh I spoke with somebody else and said I want to use AI to create products for my three-year-old daughter and maybe like a toy or something like okay so instead of being like combative it's like we were collaborative and the idea is like oh yes and what if it had this yes and what if we did the also yes and so if you do kind of the yes and philosophy you might actually just end up joining courses um worst case scenario I think there's only been one hackathon ever I've done where somebody did the exact same thing and better and and one so um you know that's always a little bit embarrassing but um it does help it's kind of screws with other people and figure out like what are the other teams doing so just in case there's like you know you have no shot of competition uh you know you're at least aware of it and I'd honestly say if like something like that happens just join forces um because you're probably gonna learn a lot uh if somebody has the same idea they thought about it for a little while and it'll help you out makes sense uh we have a question from the chat here so Aja asks how did participating in a hackathon enhance your problem solving skills slash your career and here Raul I remember you know you when we did the hackathon you were already very senior data scientist right you had a lot of data science experience so so maybe you can talk a little bit about what uh like what's made you decide to continue to do them yeah and so that's a very good point like I think I love the energy that I get when when I'm at a hackathon I love to learn uh for me learning is what winning is um and so when I go to these hackathons I am uh I'm always like looking for what people's perspectives were on a particular topic and what they are bringing to the table and learned there's no um there's there's no wrong answers in the hackathon again it's not a startup idea it's not it's not like who's going to walk away with a million dollar uh investment or something like that right it's all about um what what have you uh learned in the last few years and can take that and while contributing to your team also learn from them so that's that's maybe uh the way I see it uh yeah I I would Echo that I you know like for me I've been working on chatbots for seven eight years I went to hackathon this January I learned more about you know chat gbt than I had during the entirety of my career prior to that so hackathons are really a learning accelerator solving skills part of the question right I think it's it's about how much are you willing to acknowledge like hey you don't really like there's a different way of doing it in Engineering in general and I'm speaking from a position of like having LED teams in in companies like engineering is a team sport and it is also there's no right answers there's always trade-offs so your answer might be great but some other person has an idea you talk to them why do they want to do this in that way learn how they have attempted to solve the problem and then you just understand one different aspect of well this same problem could be solved in a different way and obviously through that you'll learn a new problem solving skill which is you know there's another way of doing it yeah um Mark asked the question says where are we going with hackathons as gen AI develops over time how do we advance the tools that Inspire new applications Caroline can you talk a little bit about that because you've done a few and I'm guessing you're going to do more yeah definitely I think that for us like hackathons really represent a awesome place for people to come together and like push the boundaries and you know a lot of these new gen AI companies are coming up from from side projects that people started out I mean I know Lane Chan I think llama index you know just started in GitHub repo so I think this yeah it's like kind of the future of how we're starting these companies that are becoming really major players in the Jedi uh landscape so I don't think we're seeing gen AI hackathons going away anytime soon I think it's just it's continuing I mean every day you look at like you look at Ben's bites or you look at any type of like AI newsletter it's like hackathon hackathon hackathon other company launches a hackathon um so yeah I think that they're I think they're important as as a way to kind of get people together get really like smart crafty like gritty people together in a room uh to build things um and that's how we're going to continue to see like large advancements in this field that's my opinion I may be biased as part of a company that's running a hackathon but I think it's a good thing um not not going away anytime soon so yeah you streamlit and llama index yeah I was gonna tip in and say well I think one of the awesome things about a hackathon is that you have to develop stuff fast right and so tools tools like llama index are very helpful tools are extremely are very helpful and and once you acknowledge like hey you don't have to reinvent the wheel well now if you have you stand on the shoulder of giants kind of a philosophy which is you have these awesome tools to help you move fast what are you doing doing after that right because you're not solving the same problem that they're solving um for example if you have wanted to build a chat interface um you know you don't have to spend time or two hours three hours building an interface you have something out there so now you can spend that time like creating some new value which nobody else has built out and so it's more like I think hackathons are a very nice um accelerating factor of innovation wherein because you're using tools it's a forcing function I'm sorry because you have time crunch it's a forcing function for you to use tools and build something new and that like create something unique oh absolutely so um we're running a little bit short on time so I I think this next question we're just going to do rapid fire here any tips or resources for improving the presentation so maybe one tip each Alex you want to go first uh yeah I posted a link to slides Carnival I'll post it again great resource for finding desks cool Raul sorry I was reading the chat I missed the question oh yeah the the question is tips or resources to improving the presentation I know you're really good at this so I I think the presentation was I I there was uh in the in the hackathon we met uh guy Kawasaki had a very good uh introduction like 10 these are these are the only 10 slides you need and somehow you know I I borrow from that and I'm like you know what let's keep the message simple there's there's a lot of stuff you can talk about but just make sure that uh you know the the Chain of Thought in the slides make sure that you win right like the the the the up at the end the last Slide the obvious conclusion for the for the judges should be yeah obviously this is this is the winner so uh that's how you go about it makes sense keep it short and to the point uh Caroline any tips for what distinguishes a good presentation versus a bad one yeah I mean I already said this but I think real world relevance real world use case like let us know how real people can use this even if it's not even if it's the more polished version of whatever you're presenting oh uh my my tip would be um have a live demo if possible if you can have a live demo that's like instant you know points in my book um okay uh saren says um and we're just gonna do this once again rapid fire style any specific problem statements any any hackathon ideas these folks are doing hackathons right now do you have an idea off your list Alex you go uh uh yeah so I think that um I'm gonna pop the list I'll be right back okay yeah I mean it's it's what inspires you right like for me the last idea I did at the hackathon was super dry for everybody else except me I built out a chatbot that helps kubernetes and like I know that I want it and so I built it and you at the end of the day if you if you really feel passionate about your idea you can also sell it like in in with your passion right I think that makes what makes a good idea is what you are passionate about and what the other person draws from you or your passion and gets excited about your idea just because you are excited about your idea oh Caroline yeah super random but I'd love to see a game in a streamlined app using AI nice my idea for you is education I've seen a lot of Education ideas win hackathons and uh I think you know it's an underexplored space for Chennai in general all right and then pulled up uh two ideas uh for anybody interested in it so uh number one is uh using large language models to generate documentation most API documentation on the web is horrendous uh there needs to be something done with that number two is either using llms or humans follow documentation and see that it actually works also a huge struggle that I have with a ton of things I think documentation is very underrated under explored yeah it makes sense good documentation really makes a big difference especially in open source um okay so the last one we're going to end on is how soon do you think gen AI codebots will win a gen AI hackathon uh we're just gonna just throw out a number go ahead Raul you first no I will say from scratch uh I don't see that happening in like maybe at least in one year but I loved like when we are doing a hackathon last month uh Alex we we were all just using chat GPT to write our our code because it was just faster at doing things in a structured way and I think it's already there if you if you count some of the recent wins it's already there it's just about does it do it itself and that's that's for you know maybe gpd5 who knows okay so not soon fully automated but sort of like semi white box is already there Alex you go yeah if you get access to code interpreter or I think it's Advanced Data analysis right now uh you can let just check to see so much work for you I agree with that cool uh Caroline what do you think I think it depends on who's judging the hackathon is AI judging the hackathon then probably pretty soon uh if not I don't know I think I think we have a while before before because when you're looking at when you're judging a project you know if it's really buggy it doesn't quite work I feel like you can use you can use AI to get you you know three quarters of the way there but without that other quarter like you're probably not winning you might get pretty close to winning so I'm gonna say at least five years okay I'm gonna go even longer I'm gonna say 20 years but if somebody at the streamlined hackathon builds a hackathon VOD that's amazing you know prove me wrong prove me wrong okay I I think you had like it's it's more like uh the cops and robbers thing right which is the the technology improves and then people are like yeah but then that you're not actually coding it you're not winning our bars are now higher right so you you kind of have to play catch up all the time exactly exactly it just just levels it up all right so on that note um we are out of time I think this has been wonderful discussion we've had some great questions um Caroline has shared the streamlined hackathon and the um Discord in the chat but it's also available um if you what is it is it just streamlined yeah it's streamlit uh dot IO slash Community slash uh llm hackathon 2023 but I will put it let me put it back in the chat real quick um let me grab this I'll just put it back in the chat there's a shortened link for you so that Link's kind of long right okay so um yeah look forward to seeing everything that you all build and thanks for joining us thank you so much to Raul and Alex and Caroline for making time this morning it was really good discussion thank you thank you thank you good luck good luck