Gitroll.io: The merit-based network for developers

GM GM. Welcome to what's going to be another fantastic year of DevNTell. So this is going to be our first session of 2025 and today I'm ecstatic to welcome Jovy Yu and Ryan Leung from Gitroll. If you didn't know, Gitroll's AI builds data-driven coding portfolios to demonstrate real software skills while eliminating tests, biases, and trivia. So if you stick around for today's episode, you'll see Jovy and Ryan showcase Gitroll's - how Gitroll highlights developer skills based on actual code, changing the way tech recruitment works today, defining devs in a dev way. All right, let's get into it.

GM GM. Welcome to the show Ryan and Jovy. Ecstatic to have you both on.

GM GM. Hey Narb.

Hello, hello.

Hello. Uh yes, welcome, you have the honor of being the first guest of 2025, which is fantastic. I love it. And I guess, usually how we like to start, I guess would you both like to give a brief introduction about yourselves before we get into the meat of the content?

Yes, great. So hey everyone, I'm Jovy. I'm from Hong Kong and I'm the co-founder of Gitroll. So I've been a tech founder for three years now and Gitroll started off as the pain point of what we faced the most in our first startup together, which Ryan and I built together as well. So we found that finding developers and knowing their actual ability, looking at regardless of where they're from and what maybe university they go to, like removing bias and finding the real builders and devs was really difficult, which is why we got to building Gitroll. We've been doing it for a year now, so we've just been helping companies find developers from all over the world.

Yeah. So hi everyone, I'm Ryan. I'm also one of the co-founders of Gitroll and I'm from Hong Kong as well. So yeah, I think Jovy has mentioned the whole story and yeah, as a dev, I'm really happy to see that I'm being part of Gitroll. I think that's a very cool experience for me and yeah, that's it for me.

Yeah, yeah, no worries, no worries. I guess, diving a little bit, how did you both kind of get into tech and development and how did you find yourselves as co-founders together?

Yes. So yeah, so like Ryan and I met in university and Ryan is a self-taught dev and we met together to build our first project which was called Metamory. It is a Web3 commemoration platform. We want to keep memories on-chain forever. So that was our first project and that's how we got into the space and that's how we got to know the Web3 environment. And through that, we were working with developers from all over the world, from India, Nigeria, but we found that a lot of times what devs tell they could do on their CV don't really match their actual skills. And that was our biggest pain point. We couldn't really find developers efficiently. And how we found our CTO Louis at the time was looking at his open-source repository. So he was a PhD student, did not have any work experience, but when we looked at his open-source contributions, he's been building and writing code since he was 13 for over a decade and that is the experience that it cannot lie. And Ryan himself is a self-taught dev, so we know that credentials don't matter as much as the real projects that a dev has built. So which is why we got together and the same team we started building Gitroll.

Yeah. I might as well mention a little bit about my backstory. So both Jovy and I actually went to business school instead of having like a tech or computer science major. So how I got into coding was I was always fascinated with building websites, I always liked doing web design. So I started off using no-code tools such as Wix and WordPress and at some point I just found these tools to be limiting and I just couldn't achieve what I wanted using plugins. So I started learning coding instead of going for an internship, I started coding learning the basics of JavaScript and React and Next.js and at that time Jovy was looking for a front-end dev, she was just starting out in her project, in the Metamory project and somehow I got on board and without any like practical work experience as a front-end dev and along this journey I've learned like Next.js and also React Native which was, as well as TypeScript actually, yeah. I learned TypeScript on the job. So I was very, very inexperienced so I was like very grateful for Jovy for having me on board during that journey. Yeah, and somehow we got here creating Gitroll.

Excellent. Yeah, sometimes people just need a chance, right? Gotta take some risk to gain that reward. Um yeah, that's excellent. So let's start to dive in a bit around Gitroll.

I know you've explained it a little bit Jovy, but did you want to give kind of the outline of what Gitroll is, why you built it, and who it is really aiming for?

Yes, of course. So Gitroll is a profile for devs to showcase their real skills based on their GitHub code. So what we do is we scan GitHub code of developers, we analyze their skill set, their tech stack, languages they've used, and we give a score and ranking towards their ability and based on how their code quality is and their contribution and influence in the open-source community. So through that, we want to showcase open-source developers and also really having a science-backed method and using our algorithm to showcase developers of their real ability instead of looking at the traditional credentials such as university degrees or using coding tests. This we see as a good pre-screening method of identifying developers and letting developers be showcased because in the way that tech recruiting is done today, a lot of amazing developers get overlooked. So this is what we want to solve.

Awesome. So yes, so this is very very apparent especially these days, right? Because of all the tech layoffs and now that the new year has rolled on, a lot of people are looking for jobs again and hiring is ramping up. So how would you say that you specifically differentiate yourselves from a traditional developer portfolio product? What's your secret sauce? What makes you different?

Yep. So what we've built, and specifically with Ryan being the lead in our algorithm and really looking into a developer's ability in a developer's way, because what we see out there in the market, a lot of the products are built for recruiters instead of the engineering managers. So we want to give a CTO perspective for empowering companies to no matter if it is the technical recruiter doing the recruiting or the technical team themselves, they are empowered as a CTO to look into developers in a developer's way. So that I would say is our secret sauce is we give the perspective of a CTO as your partner in recruiting.

Excellent. Uh yeah, I'm curious to see how this works. Did you want to bring up your screen and do a brief little demo?

Yes, of course. So of course we to show Gitroll we would love to show Narb's portfolio so we could see for everyone. So the pressure's on. So the way that Gitroll works is really simple, most every developer has a GitHub, right? So we go to your GitHub profile and get into Gitroll.io through our website and just pressing scan GitHub now. If you were to scan yourself, you can log in with your GitHub, but if you were to scan others like what we're doing right now, just press scan now and in minutes we would be able to give you a profile, a breakdown of what Narb has done in the open-source community. And this is based on your public code, but we could scan private repositories as well if it is allowed for the permission. So what we could see here is we look into the code, code quality, so reliability, security, maintainability, these are the metrics that we look into. And we also look into the influence and contribution which means the community influence on GitHub. And furthermore, what we do is we break down the languages that Narb has shown on open-source and the tech stacks that you have used, so front-end, back-end, you've got the full-stack portfolio here. And how we got into that is we look into each line of code that you've written before. So if you were to look more dig deeper into what actually we looked at, you as the CTO, like dev teams love this is really digging deep into the line of code that you've written. So everything with Gitroll is about transparency and for let's say code smells, you could be able to identify see what we detected as well. And actually some developers use Gitroll as a portfolio learning experience as well, so to track their performance over time. And for companies, what we do is we help companies discover developers from all around the world. So we help companies to tap into our talent pool. So we have over 57k developers that have signed up on Gitroll and you'll be able to use our filtering system to find the developers that would suit your needs. So basically we work with companies to find the developers that a lot of times might get overlooked.

Interesting, interesting. So that's pretty much it. So just to go into the details a bit about the scoring. So there was a score I believe that was out of 10. What what are the actual ingredients that kind of make up that score? Because you showcased the different types of things that were scanned from my GitHub, but what different parts of those scans made up my 8.76 score?

Great. So the overall score out of 10 is made up of two main parts. We call it the SMIRC. So for the hard skill code quality part is security, maintainability and reliability. That takes up around 80% of the score calculation. And we basically scan through all your code and detect if there is any bugs and errors that fit into these categories and we just assign a rank and it's also it's weighted by every repository by time and also the complexity of the project and we just take the average and that's how the hard skill part, the SMR part works. And the other 20% is based off your influence and contribution. So influence is how popular your projects are in the open-source community, how many stars you've got, how many forks etc, etc. And as for contribution is how much you have done to other projects, so how many lines of code you've committed, have you submitted any pull requests, issues, so all of that. And that makes up the 20% of the overall score. So these two components add up they add up to 10. And yeah, that's it, that's how the overall score works.

Gotcha. And in terms of the number of scans, is there I guess this kind of goes into like onboarding, if I'm a recruiter or a CTO or a company looking to hire new devs, what's the process of me onboarding onto Gitroll? Is there anything I have to do or can I just go freely paste in a bunch of different GitHub profiles into the scan?

Mhm. So right now to access our data pool, data pool of developers, companies would sign up for a company account. We will grant them access and through that we'll see an advanced version of the filtering system that I've demonstrated before. So they will be able to see over 50k profiles of developers and filter according to their needs. So let's say I'm finding a front-end developer that is S-rank in Canada, for example. We'll be able to show the profiles with the details of their personal information, their CV, experience, etc. But what we could really do for companies and companies have shown that it has worked is the pre-vetting part of the developer's ability. So we've linked quite a lot of international hires, so we have helped companies from Hong Kong find a Nigerian dev, we've helped companies from US to find a Philippine developers. We see that a lot of times with international hiring, the amount of trust that needs to be built before going to the interview process is there's a large gap there and we're hoping to fill that.

Awesome.

And is there any part of the flow where a recruiter can upload say like a job spec or a job req rather and then have Gitroll automatically match people based on that req?

Yeah, I think that we're working on it, and as well as the batch scanning as well. We're trying to kind of revamp the whole onboarding experience to better suit the companies' hiring flow needs to say. And yeah, I think we're also working on that feature where the recruiters would punch in their job descriptions and then we will automatically apply the filters for them so that it eases the process skip the whole process of tinkering with the filters because that's pretty complicated as well. So yeah, we're working on that and stay tuned. I think we're gonna do it within the next two months or so.

Alpha alpha alpha. Nice.

So that will be one of the most exciting part for Q1 this year.

There you go, you heard it here on DevNTell first. Exactly, exactly. That's what the show's for. Um so another thing that is kind of interesting peaked in my mind is since you're scanning or have the ability to get so much data about developers, is there any types of trends that you can analyze in terms of oh, many developers seem to be honing in on XYZ language or this particular framework or people seem to be making the same type of code smell mistakes? Is there any type of like high-level data analysis available or some trends that you've seen you might want to highlight?

Well, for me, I think it's because we supported JavaScript and Python earlier and we added like more languages down the road which is C and Java. So most of our data actually comes from JS developers. And one of the trends that we observed is that React is still the king but then other smaller frameworks like Astro and Svelte is actually quite popular as well from a lot of front-end devs actually write in Svelte. And as for Python, yeah because of the AI trends, a lot people are writing more in Python, Jupyter notebooks and that's on the rise as well. And we see that many devs have also at least have one kind of Jupyter notebook project on their GitHub. Yeah, so that's kind of the trend that we have observed recently.

Yes. And to add on that, we've been also seeing more devs that are getting into Web3. So in particular, we have a large pool of devs from Africa because we worked with one of the largest tech influencers, Sultan, in the African space. And we found that actually a lot of the developers from Africa are building on the Web3 space, which is pretty exciting. And we also did a collaboration with Chainlink last year and we saw like overwhelming support in the community that are actually attending the conference in person in Hong Kong. So we do see like Web3 trends are on the rise within our community.

Yeah, yeah, I'm sure they'll be rampant this year as well with all the crypto AI agent frenzy that's going on. Um but yeah, that's excellent to see. So I'm curious, like what was the what was the biggest challenge you guys faced while building Gitroll? I'm sure there were quite a few, but was there anything in particular that stood out as like oh, we got over this particular thing that really like got everything really rolling?

I think for me is how to the two parts mainly. I think the first part is how do we get the algorithm to be not so CPU intensive, not so computing power intensive because like we the way that we did at the most we scanned every single repository and that took ages like some repository are really really big and it actually crashed our server. So we had to add a little bit more limitations, a little bit more workaround to make the whole thing a little bit lighter and to keep and we try to maintain as much detail as possible and have a reasonable scan time. So that was one of the challenges that we faced between the accuracy or the comprehensiveness of the scan and also the performance. And the second thing is of course adding new languages. So for each language that we add we had to take a look at all the rules to make sure that we're not getting a lot of false positive rules and that's a lot of manual labor and it's very time consuming as well. But yeah, at the end we've added C and Java and it took us three months to add both languages and it was really a huge task but yeah, more languages are on the way and yeah I'm happy that these two languages which are very popular are finally implemented into Gitroll.

Awesome. Yeah, I can imagine the difficulty.

And is there in any place in the stack where you're using AI to kind of help alleviate some of that manual any type of manual process?

Yeah. Yeah, we used AI to help us build the dependency database, we call that. It's to how to detect the libraries and also and to show it in the tech stacks page that we've shown in the Gitroll profile. So that was with the help of AI and also we've used AI to kind of help us identify the complexity of the project. So sometimes we encounter a lot of like kind of toy projects where they just create React app and then just wrote a Hello World and that type of project it's gonna take us a lot of precious processing power to scan these. So using AI to kind of identify these projects and skip them. And this help reduce our costs as well.

Nice. Great to hear. And I guess as AI is one of the one of the leading trends of last year and probably this year as well, and with the rise of AI code editors, do you foresee potentially like any issues where people are people have these code bases but maybe not necessarily built by themselves, like maybe it's cooked up by an AI and might not give the full context of the actual developer's skills? Are you guys working on ways to be able to detect whether a code base was generated by AI or any safeguards in place for that or at least are you thinking about that?

Yeah, we're thinking about that. Right now we think that AI still has its limitations in terms of the complexity of the project I think. So AI is really good at creating very simple websites and snake game of course like all the AI code writer which replaces the actual developer, they also like to use snake game to demo. Right now what we're doing is we actually take a look into the complexity of the project, the code complexity and compare it with the lines of code. And if it doesn't meet a specific threshold, we would deem it as an AI project written by an AI most likely. It's still it's not the best way of doing that but I think it's a simple way workaround it works at the moment kind of. But yeah, I think with the rise of AI more and more developers are writing code using AI, which is good I think if you could use AI to assist you in writing code, that's also part of your ability as well. And yeah, I think we in our company we also embrace the use of AI. I think it's inevitable and we actually encourage our devs use AI as well. But at the end it's about how the how you use AI and is the code clean.

Yeah. Yeah, definitely, definitely. It's about the prompt engineering and then verifying that the thing that it spit out is actually what you want and correct, 100%, 100%. Uh yeah, that's awesome to hear your company is embracing that as well. I think we'll see more and more of that as we keep going on. Anytime you can get any time back or accelerate your work, I think that's obvious win.

As we're approaching time, time flew by, didn't it? Just want to kind of bring it back to the audience in terms of how they can get started with Gitroll and if they if people and companies are interested in working with Gitroll, how can they get in touch, what's the best way to get started?

Mhm. So for all the developers could just go on Gitroll.io to be build your own profile, see how your score is, tag your friends as well, it's a great friendly good competition we've seen devs around the world have been doing. So yeah, it will be a good thing to do and for companies so we have the Find Dev Gitroll.io Find Dev page and through that you can get in touch with our team and we would help you getting access to our talent pool and helping you find your next 10x dev no matter where they are.

Awesome. Love it. Love it. You heard it here folks, definitely get on that, scan your own profile, why not share it on social media too, tag Gitroll, see where you stack up. But just want to thank you both Ryan and Jovy, I really appreciate both of you taking the time to come on today and give us this excellent overview and kind of go in depth on Gitroll and how you both kind of started with your team and the company. It was awesome, awesome insights.

Yeah, thanks Narb, honor to be here.

Yes, great, great to start the year with.

Exactly, exactly. Let's just get it going, let's get it going. Exactly. And just before everyone takes off, just wanted to highlight that we've begun on the show to partner with BitBadges to give or to enable viewers of the show who view us live to claim an attendance badge. So if you didn't know, BitBadges is created by DD member named Trevor Miller, previous guest of the show, and BitBadges allows you to collect multi-chain NFT badges, earn credentials and build your universal digital identity. So what you'll want to do is scan this code, sorry for blocking your face there Jovy, what you want to do is block scan this code and fill out a simple form to claim your Gitroll attendee badge for viewing us live on DevNTell today. And you'll have one hour from now essentially to be able to claim that. So definitely get on that if you are into that the claiming collectibles game. But with that, just want to wish everybody a very happy Friday, happy weekend and we will catch you back here next week for another episode of DevNTell. Cheers.

Cheers.

Cheers.