Talent from Rural Ethiopia - Markos Muche, AI Scientist at iCardio.ai

Talent from Rural Ethiopia - Markos Muche, AI Scientist at iCardio.ai

From climate change to cardiology, 25-year-old Markos Muche has made it his mission to utilize AI towards bettering his native Ethiopia as well as communities around the world.

Born to a farming family in the rural Gondar region, Muche is now one of his home country's most promising young engineers, having received the top prize in Ethiopia's most renowned AI competition. Prior to this, he had beaten out 200,000 applicants to get the highest possible score on Ethiopia University's entrance exam.

‍

Today, Muche is a Machine Learning (ML) engineer at iCardio.ai. There, he works under Chief Scientist, Dr. Roman Sandler, to train the company's advanced deep learning algorithms to interpret ultrasounds in lieu of doctors. 

The position is highly important at iCardio.ai, an innovative start-up disrupting the field of echocardiography by pairing machine learning algorithms with an unprecedented dataset, comprising over 200 million individual ultrasound images. Muche is tasked with "teaching" computers to recognize patterns and become more accurate at interpreting outcomes without being explicitly programmed to do so. 

“Basically, there’s an algorithm. First it accepts an image from the machine and the module takes that image and classifies it and you write an algorithm, a code, that checks if it’s correct,” explains Muche. 

“If it’s correct there will be a type of reward and the module will understand if it’s done well and has less loss, otherwise it will self-adjust,” Muche continues. “When it makes many or a big mistake, it makes self-adjustments to its parameters to avoid future errors.”

Making Muche’s success particularly impressive is the fact that he is entirely self-taught. Getting to such a prominent role required years of hard work that saw Muche singlehandedly teach himself Artificial Intelligence, Machine Learning and the complicated math and data science underpinning it.

His interest in the field began when he was first exposed to the basics while in university, leading Muche to spend the next few years learning everything there was to know. Amongst other things, Muche taught himself Pytorch and Tensorflow, libraries built for machine learning, in order to pursue publishing research around a specific field of machine learning called transfer learning. 

From there, he delved into the world of deep learning, an element of data science that uses machine learning and artificial intelligence (AI) to imitate how humans gain new types of knowledge. The journey required him to comb the internet for free resources along with hours of daily studying in order to get to where he is today.

“I studied AI day and night. When I saw a video or read an article, it was about AI,” relates Muche. “When you immerse yourself in it and when it’s not a struggle, because if you don’t like something it’s a struggle to learn it, but since I like it it's not a struggle. There’s no friction.”

His effort paid off; within a few years, Muche won first prize at a University of Addis Ababa on using AI to fight climate change. In what he calls “AI for agricultural monitoring”, Muche and his team found a way to reduce the amount of pesticides sprayed on fruit trees and plants by using AI to analyze photos of the said orchards. 

“I got more immersed into it because it interested me and still interests me. That’s how I started and then I started learning AI from a lot of online resources,” Muche recalls. 

“I found deep learning to be an incredibly profound and extensive concept. I found myself discussing deep learning with my friends at school,” Muche relates. “I am so inspired by Neural Link, where we one day may be able to communicate with our brains, just thinking! I understood that deep learning is the future.”