Vérité Mugabo Makuza, 19, has developed Kobra, a visual programming tool designed to make machine learning (ML) easy to learn and experiment with.
The graduate from Rwanda Coding Academy (RCA) recently secured USD1 million from OSS Capital, a firm based in San Francisco, USA, for the project.
Machine learning, according to TechTarget Network, is a type of artificial intelligence that allows software applications to become more accurate at predicting outcomes without being explicitly programmed to do so.
Recommendation engines such as YouTube, Netflix and Amazon are a common use case for machine learning. Other popular uses include fraud detection, spam filtering, malware threat detection, business process automation and predictive maintenance.
A snapshot of Kobra user interface. Photo: Courtesy.
“Kobra helps people who do not know much about programming to understand machine learning,” said Makuza. “If you can understand English, you can understand my project. You just go to the ‘editor’ section and it directs you to do and achieve what you want. It allows you to learn and apply what you’ve learned.”
Makuza started developing the program in 2020 when he was in second year at RCA. He joined the school in 2019 without any programming skill.
According to him, the school helped him to unveil his potential as he explored more about programming, working on various projects.
When an idea of making something that could help people understand machine learning came to his mind, he found it interesting given that he had realised that computers are very intelligent devices people take for granted.
“I wanted to show people how intelligent and useful computers are by empowering them with the right skills to know that,” he said.
He started building his program “Kobra” from scratch and uploaded its prototype on Github, a code-hosting platform for version control and collaboration.
When he shared it, people loved it and compelled him to put more effort, with some suggesting adding significant features to improve the program.
“It gave me much satisfaction to realise that what I had created with my brain and hands was actually useful to the world,” he said.
Talking about how Kobra works, he said, “A user can try it without logging in but can opt to log in to access more features. A user can upload a dataset, visualise it by using different graphs and use it to train a machine learning model.”
A machine learning model, according to Microsoft, is a file that has been trained to recognise certain types of patterns. Usually one trains a model over a set of data, providing it with a set of rules it can use to reason over and learn from those data.
On the other hand, a data set is a collection of related, discrete items of related data that may be accessed individually or in combination, or managed as a whole entity.
Investment potential and future plans
Tackling how he met Replit Ventures, Makuza said he sent them a demo of the project and they loved it and posted its video on their Twitter handle. When the investor, OSS Capital, saw it, they contacted them and connected with Makuza.
Makuza is currently developing a robust plan on how he will use the money to upscale his project and raise seed funds.
To him, the reason why the investor got interested is because the skills of machine learning are most of the time attained at Masters and PhD level and not common in the tech world.
Apart from that, applications that include ML are also useful because they provide accurate information to users and provide better experiences, and thus having a program that helps people to learn the skill in an easy way is “impressive.”
“There are many people who want to acquire ML skills but good sources of learning are rare. Some even feel like it’s hard because at some point, it involves mathematics or that it requires one to have a huge background in programming which is not true, because if you focus on something, you can understand it,” he said.
Makuza also believes that his achievement is going to help people who have interest in technology to make more interesting inventions, urging the public to explore many innovative products.
He plans to add a feature that allows other developers to make applications by using Kobra and to scale up the program to reach a bigger user base which is currently 1000 maximum.
“I also plan to translate the content of the program in Kinyarwanda so that Rwandans can easily learn the skill,” he continued, adding that he is also set to join university to sharpen his skills further.
To learn more about how Kobra works, you can check out this video.