AfriCan: Celebrating the Tunga AI Hackathon Winner!

“Our team name AfriCan was coined from African-Can. We believe African developers can also build amazing things in the AI space” 

• Paul, a member of the winning team.

Paul is an Electronics and Electrical Engineering student at Obafemi Awolowo University and a Machine Learning Ops Engineer —- He was excited to share his hackathon experience with the Tunga team as his team emerged the winner of the Tunga AI Hackathon held on the 13th to 17th November 2023, hosted by the Tunga community. This is an inspiring story of 3 African Artificial Intelligence and Machine Learning Engineers who believe that AfriCan.

Tunga AI Hackathon: An Event like no other

In a world driven by technological innovation, coding is the universal language of progress. It’s the key that unlocks doors to a future limited only by our imagination —- Tunga wants to be at the center of the innovation by spotlighting the amazing developers in Africa, we believe in the power of code and its infinite possibilities. That’s why we organized the Tunga AI Hackathon-themed “AI Solutions to Help Developers Code Better”, an event designed to propel the coding community into the future of the AI era.

We are excited about the past couple of weeks —- after the announcement of the Tunga AI hackathon we received over 250 registrations across Africa, 15 teams were formed from these, and 12 phenomenal projects were submitted. We believe this is a step in the right direction for African AI engineers.

“I am not capable of judging the technical quality of the projects.  Nevertheless, I was proud and humbled by the participants’ efforts and dedication to participate in our initiative. It’s always special as an entrepreneur to realize that some crazy idea you once had has led to a chain of events that touches people’s lives in ways you could never have imagined whether they win or not, it clearly benefits them to participate, I’m grateful to everyone that made this hackathon possible” 

Ernesto the CEO at Tunga, says these words with a smile on his face during the project presentation of all teams. His heart swelled with pride as he saw the amazing project ideas presented by these talented individuals. He then concluded by addressing the participants of our hackathon, saying that they are already all winners because by building AI technology skills, they make themselves attractive to international tech companies.

Meet the AfriCan

The emerging team consists of 3 male Electronics and Electrical Engineering students of the Obafemi Awolowo University, Ile-Ife Nigeria in the western region of Africa. Paul is the team lead and Victor and Dunsi are the supporting strikers. They are all driven by their passion for technology and didn’t let their being in school deprive them of the opportunity to create an amazing AI tool that helps developers code better.  This group is a born winner as they have won different hackathons in the time past. We had the opportunity to interview them and here are their responses;

Q: Congratulations once again guys on your win, did you expect it or were you like; “well…on God”?

Paul: Funny thing is we knew our project was a really solid one, we knew it stood a chance of winning but we weren’t just sure of the way the judges would see it. We also had similar projects with other participants. And yeah, we were confident initially before the presentation but after the presentation, when we saw others, I think we started to second-guess our project.

Victor: The first presentation was like a VS code extension. I was like ah😯we thought that was the unique thing about our project. Yeah, it was just maybe how properly we implemented all the features and the usability.

Q: How did you and your team hear about the hackathon?

Paul: Yeah, so the funny thing was I saw it on Twitter once then, I saw it on Gift’s WhatsApp status because I happen to have Gift’s contact. And that was when I saw it the second time and I thought: okay, what is this hackathon even about? So I checked the details. Then I, Dunsin, and Prof. Oh, I call Victor Prof. 😁, we have a group. So immediately I saw it and I checked. I sent the link to the group and I was like, “Guys, shey make we run am?” [i.e. Guys, should we participate?]. So we saw the timeline and we saw seven weeks and I’m like okay, that’s a long one. But I got to understand that the building itself is not till the seventh week, there were going to be some extra events. So we decided we were going to go about building.

Q: Interesting. Okay, so basically you heard through one of the mentors. So was Gift your mentor?

Paul: No, our mentor was Blessing Adesiji.

Q: I know you just mentioned that you guys are on a WhatsApp group already. So you guys are like brothers from another mother, right?😁 How did you guys meet? 

Paul: Something like that, we are the hackathon brothers😂

Victor: If you say so.😄 Okay, so I’d say oh, how we met is a very complicated question but first, we attend the same school and that’s Obafemi Awolowo University.

Q: Are you guys still students? 

Victor: Something like that. So Paul and Dunsin are still students. They are currently in their final year. I’m also in the same department, we’re all in Electronics and Electrical Engineering. But I just finished so I am a year ahead of them. So we have this DS club in school, (Data Science club) and Paul currently leads the club. We are very active members of that club —- we got to connect and we started doing hackathons together since, 2020. Is it 2020? I guess…Yeah. So I think that’s pretty much what I can remember.

Paul: So, basically, we got to know each other right from the community. We organize events together. We attend events together. We also do hackathons together, and we happened to have won a lot. One hackathon we won in the past is the very popular UmojaHack Africa. We won UmojaHack Africa consecutively, for three years. We just tend to do hackathons together because we know each other’s strengths and weaknesses and we just know how to balance them.

Q: It’s really interesting that you guys have been winning hackathons before now. So you guys are born winners 🤩 What are your specializations? You know, are you into Data Science, ML? Are you just a Full Stack? 

Paul: So for me yeah, I started out doing just Data Science. Then I moved on to doing purely ML. Then, I got into doing ML Ops. ML Ops is like the aspect of ML that helps you to deploy models. Everything that comes after building models. So then given this current wave of Generative AI, I upskilled to learn just the basics of using Open AI and other open-source LLM providers that’s it. So lately, I’m just doing LLM stuff and ML Ops.

Victor: Yeah, for me as well. Pretty much the same thing. I do Data Science and Machine Learning, essentially. And I do a lot of competitive programming. So I’m skilled towards that line, [working with] platforms like Kaggle if you know those platforms. Also, I pivoted into NLP sometime last year, and I also did some sort of ethical AI stuff. But now I’m pretty much you know, in the space of NLP and most recently, LLM, just like Paul.

Dunsin: For me, I started with Python then I moved to NLP. Normally,  I’ve already stayed [alone] in the industry, but Paul and Victor just joined us😅So I do NLP majorly. I [also] specialize in LLM and Generative AI.

Q: That’s nice. The next question would be, can you tell us a bit about your solution, like in brief, you know, just a description of it, and the major features? 

Paul: Okay, so our solution is an AI-powered VS code extension that helps you to generate code, fix errors, generate documentation, and also unit tests. These are the four main pain points for developers. Regardless of your journey, you could be a beginner, an intermediate, or an expert and these are like still the common issues we tend to have. [Although] there’s also the part of chat GPT. Given the rise of the likes of chat GPT and asking it about stuff, then it gives you some code. You copy your code and try it out in your IDE. Then you run it. Sometimes it can hallucinate and give you things that are not real. So you run into errors. Developers have to copy errors again, give it to chat GPT to give you something better, and then come back. So [there is] this back and forth. Why not have this tool built into your IDE such that you only have to prompt it to this stuff? It does it and If there’s an error, you fix it right there. If you need to explain errors, you have error explanations in the comments — So it gives you a brief explanation and all. It also helps you to do the tedious work of documenting code and testing code that is unit tests. ‘Don’t know if that was summary enough.

Q: Yes, yes it is but I’d like to ask a follow-up question. I know you said some things about ChatGPT and its flaws but what other thing inspired your solution? You can also share a personal experience or what you know other developers in the community have faced. 

Paul: Okay, I can give my personal experiences and pass [the question] on to my team members. So personally and I think generally, I enjoy reading stuff. But sometimes I can get discouraged when I do too much planning before even doing the main thing which can slow down things. So there is an approach that can help you do something way faster, where if you already have a head start, you tend to follow through with what you want to do. 

So in this case, now you have an idea [and] you want to build a website for your blog. You probably may not be too proficient in HTML, CSS, and the like but if you can just prompt a tool, say there is a tool that can help you to generate a code to get started. And you just have to follow. Since you have a code to start with, you just continue working and all — So one of the things that inspired this, is you have your IDE where you are coding and you have an AI assistant right there and it’s helping you to achieve what you want to achieve. 

Victor: So for the hackathon, Paul essentially brought most of the idea. In terms of my own experience, as an ML Ops, I can completely relate to those problems; writing docs, writing code, Debugging, etc. The tools essentially help write better code.

Dunsin: Yes, so a lot of times when I’m coding and I get errors, I tend to copy the errors inside the terminal and paste it on Google a lot of time it’s usually frustrating searching for errors. And it was more like okay, if there is a tool that can actually just explain an error instead of going from one tab to another so that it will be easy. Whenever I paste code inside Google, Google does not understand the context from which this code is coming. I only paste the error, I can’t paste the whole code on Google search. So I mean, it makes more sense that fixing errors and explaining errors come from the VS code terminal. So I loved that something like this could be done so I tried to work on this. Then, there’s generating code documentation. So, a lot of time, I don’t put README on code and when I get back to it, I will be like “Where do I start from?” — so it just makes things easier. When we have something like documentation even I will be able to understand what I’ve written in the past. So I think that was the main inspiration for this project.

Q: Very nice. Thank you so much for sharing that. Okay, this next question, I know most people, especially other participants in the hackathon will want to know this. Did you start working on this project during the hackathon or have you been working on it before then?

Paul: So given that you said we had seven weeks, seven weeks was what I saw on the website but it turned out that we only have a few days to build. So the idea was for everything we want to implement to be documented before the building starts. So we were doing a lot of brainstorming in Google Docs, about everything we wanted to implement, how we wanted to sell the idea, how we wanted to write and build the VS code extension, and other things, everything required for us to build was done before the building time —  and we weren’t sure if we were supposed to start on that day and not before. So that was why we decided to just wait and do the brainstorming. And once it was time to start building, it was easier for us to piece things together. And we built faster given that we already had everything we wanted to do and what we wanted to add to the slides. So that may have made it easier for us given that we knew the timeline and we just worked around the timeline.

Q: So basically, the actual building started when the hackathon itself started. I mean during the five-day hackathon?

Paul: Yes we did. A funny story was that two days before the submission we thought we were all set because some features were working and it felt okay given these features are working we can make a video or we can make a demo around it. But it turned out that we started getting errors and bugs. We exhausted the credit on the API key. And we didn’t know how to give the ID a new key to start working through the prompts, when we put in the key it was supposed to give us the prompt again but we started getting errors and it was just a little challenging towards the end and yeah, we just had to work with what we had and it happened that we submitted the final demo and everything just some minutes to 12. That was close to the deadline. 

Q: Nice. I was also going to ask about what the experience was like working on the solution within a few days. So, I hear you answer that a little bit. If you want to add something else about the time constraints and all of that…

Paul: Nothing more to add. Given that I and Dunsin are still very much in school. So we had schoolwork and we also had the goal of winning the money so we had to fight between: “Do you want to write code or do you want to read the books? So we tried to balance both. Happy it turned out well. We didn’t lose the books and we’re also able to win the money. 🤭

Q: Nice, nice. Now that you have your solution working. I know you made it open source right? Do you have any goals that you want to achieve with it?

Victor: Yeah, the solutions are open-source and available on GitHub. And also, because we also wanted ease of access to use. So if you look at the way the solution is built. We already have the VSS file, instead of compiling from GitHub directly. So essentially, like maybe we could potentially release you know, an official VS Code extension. Who knows?😁😁

Paul: Yeah, something I want to add. So on our GitHub repo, which is public, by the way, there’s a VSS file that anyone who wants to use it just has to clone the repo, and then install that file. Just like we have an APK or stuff like that. But the way very standard and finished extensions are used is that you just go to your VS code IDE, search for the extension, and install the extension. So one of the things that we like to be our next step is to make it installable just by searching for it on the VS code extension. Although there are still some improvements to make…it’s not perfect, yet. It does everything you want it to do but it’s not perfect because in applications where you need to use prompting some things can go wrong. And yes. Sometimes you want to do multiple things at the same time. You may not get what you want. So, say you want to generate README documentation and generate unit tests at the same time, there may be some issues. Also, there are minor things that we feel like we can work on to make this a very standard VS code extension and for people to contribute to making it better.

Q: So do you have a timeline in mind, to fine-tune all of these things and maybe create an extension?

Paul: Okay, given that we are not constrained to just five days to do that again. What we want to do is try to see how we can balance building it with what we were doing before the hackathon. So for me and Dunsin, it is mostly school. Victor probably has something he was doing before. So we will make sure that the project doesn’t stop and at the same time try to improve it.  I don’t want to give you a strict timeline, but let’s say a month to 2 months and it will be installable by searching on the VS Code extension.

Q: So that’s nice. Aside from what you’ve said, are there any next steps for you and your team? You know, personally, professionally, you know…whatever.

Paul: So I guess mostly next step would be adoption. But the issue is if it’s with developers adopting and using this extension and gaining users to use it, there’s a constraint of using an API key. And if you want to get an API key, like API credit, which is not free, the issue is…Do we want to make it paid? Or do we want to just leave it to people only who have API keys? So there’s a decision [to be made]. If it’s going to be used by people how do we want to solve this problem, where you need an API key, and not everyone has an API key? And is it good enough for people to pay you to use it? These are conversations we’ll be considering going forward.

Q: So basically are considering monetizing

Paul: It’s monetizing because it’s expensive to run it. You need API credits to use it. How are you going to keep buying credit? So that’s the dilemma. I know we will figure it out when the time comes.

Q: So an interesting question, if you don’t mind us asking, how do you and your team plan to spend your prize money?

Paul: I’m going to let Victor and Dunsin go first😃😃

Victor: I have not thought about it. I might just save it

Paul: I’ve some gadgets in my cart. I’d probably just buy them. I’ve seen free money.😅

Dunsin: Apart from the personal things I will get, I will spend it on research. 

Q: What are your thoughts on the future of AI in the tech industry?

Victor: Hmmm, this one is deep. For me, there has been a buzz around generative AI recently and of course, the whole AI wave coming from far back when computers were still struggling to recognize images to where we are now. There has always been the wave of oh this is the next big thing and everybody jumps on the train —- the same thing happened with deep learning and it’s happening with generative AI now. I think the space will continue to innovate and we are getting to this future synergy where everybody uses AI not just the elite or the developers. I see a future where everybody can create their own AI. Creation will be easy, adoption will be easy. I see integration, adoption, and ease of doing things. I will just stop here so I don’t give a lecture😃

Q: I see you are very good with lectures😁 so Paul do you have anything to add?

Paul: I think Victor has said it all. I see a future of more adoption, we need more people in the space. Instead of people having that fear, they just get to roll with it. AI is not the hardest thing as it’s been portrayed. Anybody can be an AI Engineer. We also need more communities like Tunga to embrace the development and adoption of AI. As a developer, AI makes me stay at the top of my game because I feel challenged that what I can do, AI can also be programmed to do the same. I don’t feel threatened, instead, I feel challenged to always want to do more. We can use the existing technologies to build more solutions for ourselves — things that are unique to us Africans because no one will do it for us.

Victor: I see the African AI tech ecosystem taking shape and I believe in what we can do in the next five to ten years.

Paul: I don’t know if you got the hint from our team name “AfriCan” like “Africa Can” you gerrit?😁😁😁

Q: Have you guys joined the Tunga community?

Paul: We hadn’t heard of Tunga before the hackathon. I went to research Tunga and understood what you do after joining the hackathon. We have seen benefit of being part of the community so we will definitely join.

Q: We will be launching an AI open-source project next year, is this something you are open to?

Paul: Yes, definitely. We would like to join the community and contribute to the AI open-source project. For us, when we are building we leveraged on a lot of AI open-source projects. It just makes more sense that we also make our project open-source.

Q: I just want to say I’m excited and I’m proud of you all.

Paul: Thank you.

Victor: Thank you

Dunsin: Thank you

And there you have heard it from the horses’ mouth. This interview was filled with laughter and insight because these individuals know their onions. We want to wish the AfriCan team a big congratulations. It’s time to spend some euros🤩

Tunga’s Mission

And there you have heard it from the horses’ mouth. This interview was filled with laughter and insight because these individuals know their onions. We want to wish the AfriCan team a big congratulations. It’s time to spend some euros🤩

Tunga’s mission is to create 21st-century jobs from African youths and provide opportunites for African tech professionals by giving access to the international tech scene.

As a software developer, AI engineer, Machine Learning Ops, or mobile developer, you can be part of this growing revolution in the AI African tech space if you — just as the AfriCan team did — choose to capture the moment and never let it slip.

Tunga wants to take you further. Join Tunga by clicking on the link.