As part of the joint project by Digital Business and Astana Hub, “100 Startup Stories of Central Asia,” Najima explained why she didn’t enter IT right away, how epsilon3.ai came about, and how the company’s solution can prevent large-scale electricity theft. She also shared how difficult it was to negotiate implementation with government agencies and why the project is not currently considering expansion into European markets.
“I Was Convinced That IT Was Only for Men”
— Najima, how did your journey in IT begin?
— In 2018, I graduated from school and faced the choice of where to apply. Even though I studied in a mathematics-focused class, I was convinced that IT was a male-dominated field and that I didn’t belong there. So I enrolled at the Russian State Social University in Moscow, majoring in Sociology.
During my studies, I constantly took on side jobs, and in my fourth year I decided to look for a full-time position. I managed to get a job as an IT recruiter at an agency. In the hiring process, I mostly communicated with frontend and backend developers, but I also interacted with ML engineers, AI product managers, and other specialists working with artificial intelligence. I was still convinced that AI wasn’t for me, but I realized I wanted to be on the other side of IT.
After graduating, I returned to Tajikistan and switched to remote work. At some point, I realized it was time for a change. I started taking product management courses and looking for an on-site position in Dushanbe — remote work didn’t suit me.
— Which roles were you considering?
— I applied to many different vacancies, but it felt like my place was somewhere within international organizations. A close friend of mine had experience working with zypl.ai and suggested I try applying there. I chose a random position at the company and sent in my resume.
HR contacted me. I explained that I didn’t have much hands-on IT experience, but that I really wanted to be part of that team. In the end, I was offered two roles: office manager or CEO assistant. Naturally, I chose the latter and worked as an assistant to Azizjon Azimi for about a year.
During that time, I learned how a startup works from the inside and significantly developed my skills. In the spring of last year, when the A7Sigma holding began to take shape, I was appointed manager of the A7σ executive office. Later, I was promoted to director. During this process, the idea for epsilon3.ai emerged, and it began to be developed by the A7σ team.
“Our Models Achieve Up to 99% Accuracy”
— What does epsilon3.ai offer today?
— We focus on analytics and forecasting for GovTech companies using zGAN, a synthetic data generator developed by the R&D team at zypl.ai. This technology belongs to the GAN (Generative Adversarial Network) family — models based on game theory principles. A GAN consists of two components: a generator that creates synthetic data closely resembling real data, and a discriminator that learns to distinguish real data from generated data.
What differentiates zGAN from similar networks is its ability to deliberately generate outliers — anomalous data points that significantly differ from the rest of the dataset. In traditional machine learning, such outliers are usually removed because they can distort the model. We do the opposite: by adding synthetic outliers, we identify deviations from the norm. For example, in the case of an energy company, zGAN can indicate where unexpected events may occur — a sharp spike in consumption, a sudden drop, or energy usage when meters show zero.
In one case, we analyzed the Uzbek market. According to official data for 2024, electricity theft in the country amounted to $108 million. Even if our system can detect just 20–30% of that volume, it would save tens of millions of dollars.
— So epsilon3.ai and zypl.ai are built on the same technology. Why create a separate startup?
— zGAN was originally developed for fintech use cases. However, as the technology evolved, it became clear that its potential extended beyond finance and could be applied in other sectors, including government digital services. At the same time, zypl.ai remains focused on fintech solutions and could not actively scale work in the GovTech direction. That’s why we decided to spin off a separate project for new industries.
— How did GovTech companies respond to your product?
— Approaching cold clients usually leads to strong resistance. So we chose a different strategy and actively leveraged the opportunities provided by the broader Central Asian IT ecosystem.
At the end of 2024, we applied to the Industrial AI Accelerator, supported by the Ministry of Digital Development and Aerospace Industry of Kazakhstan. This program brings together businesses and government institutions to implement joint use cases. That’s where we found our first client — the Kazakh company Karabatan Utility Solutions, for which we developed a pilot project.
After that, clients began approaching us with concrete requests. In addition, many potential customers are already familiar with zypl.ai’s work. Mentioning the company strengthens trust thanks to its recognized expertise in artificial intelligence and international fintech experience, which gives us a certain level of credibility.
Astana Hub also plays an important role. For example, with their support over the past few months, we presented the project to the President of Kazakhstan, the King of Jordan, and the Mayor of Dushanbe. I am physically based in Astana, so many inquiries come directly. For instance, we recently received an invitation to meet with the Ministry of Energy of Kazakhstan.
Read more on Digitalbusiness.kz.
As part of the joint project by Digital Business and Astana Hub, “100 Startup Stories of Central Asia,” Najima explained why she didn’t enter IT right away, how epsilon3.ai came about, and how the company’s solution can prevent large-scale electricity theft. She also shared how difficult it was to negotiate implementation with government agencies and why the project is not currently considering expansion into European markets.
“I Was Convinced That IT Was Only for Men”
— Najima, how did your journey in IT begin?
— In 2018, I graduated from school and faced the choice of where to apply. Even though I studied in a mathematics-focused class, I was convinced that IT was a male-dominated field and that I didn’t belong there. So I enrolled at the Russian State Social University in Moscow, majoring in Sociology.
During my studies, I constantly took on side jobs, and in my fourth year I decided to look for a full-time position. I managed to get a job as an IT recruiter at an agency. In the hiring process, I mostly communicated with frontend and backend developers, but I also interacted with ML engineers, AI product managers, and other specialists working with artificial intelligence. I was still convinced that AI wasn’t for me, but I realized I wanted to be on the other side of IT.
After graduating, I returned to Tajikistan and switched to remote work. At some point, I realized it was time for a change. I started taking product management courses and looking for an on-site position in Dushanbe — remote work didn’t suit me.
— Which roles were you considering?
— I applied to many different vacancies, but it felt like my place was somewhere within international organizations. A close friend of mine had experience working with zypl.ai and suggested I try applying there. I chose a random position at the company and sent in my resume.
HR contacted me. I explained that I didn’t have much hands-on IT experience, but that I really wanted to be part of that team. In the end, I was offered two roles: office manager or CEO assistant. Naturally, I chose the latter and worked as an assistant to Azizjon Azimi for about a year.
During that time, I learned how a startup works from the inside and significantly developed my skills. In the spring of last year, when the A7Sigma holding began to take shape, I was appointed manager of the A7σ executive office. Later, I was promoted to director. During this process, the idea for epsilon3.ai emerged, and it began to be developed by the A7σ team.
“Our Models Achieve Up to 99% Accuracy”
— What does epsilon3.ai offer today?
— We focus on analytics and forecasting for GovTech companies using zGAN, a synthetic data generator developed by the R&D team at zypl.ai. This technology belongs to the GAN (Generative Adversarial Network) family — models based on game theory principles. A GAN consists of two components: a generator that creates synthetic data closely resembling real data, and a discriminator that learns to distinguish real data from generated data.
What differentiates zGAN from similar networks is its ability to deliberately generate outliers — anomalous data points that significantly differ from the rest of the dataset. In traditional machine learning, such outliers are usually removed because they can distort the model. We do the opposite: by adding synthetic outliers, we identify deviations from the norm. For example, in the case of an energy company, zGAN can indicate where unexpected events may occur — a sharp spike in consumption, a sudden drop, or energy usage when meters show zero.
In one case, we analyzed the Uzbek market. According to official data for 2024, electricity theft in the country amounted to $108 million. Even if our system can detect just 20–30% of that volume, it would save tens of millions of dollars.
— So epsilon3.ai and zypl.ai are built on the same technology. Why create a separate startup?
— zGAN was originally developed for fintech use cases. However, as the technology evolved, it became clear that its potential extended beyond finance and could be applied in other sectors, including government digital services. At the same time, zypl.ai remains focused on fintech solutions and could not actively scale work in the GovTech direction. That’s why we decided to spin off a separate project for new industries.
— How did GovTech companies respond to your product?
— Approaching cold clients usually leads to strong resistance. So we chose a different strategy and actively leveraged the opportunities provided by the broader Central Asian IT ecosystem.
At the end of 2024, we applied to the Industrial AI Accelerator, supported by the Ministry of Digital Development and Aerospace Industry of Kazakhstan. This program brings together businesses and government institutions to implement joint use cases. That’s where we found our first client — the Kazakh company Karabatan Utility Solutions, for which we developed a pilot project.
After that, clients began approaching us with concrete requests. In addition, many potential customers are already familiar with zypl.ai’s work. Mentioning the company strengthens trust thanks to its recognized expertise in artificial intelligence and international fintech experience, which gives us a certain level of credibility.
Astana Hub also plays an important role. For example, with their support over the past few months, we presented the project to the President of Kazakhstan, the King of Jordan, and the Mayor of Dushanbe. I am physically based in Astana, so many inquiries come directly. For instance, we recently received an invitation to meet with the Ministry of Energy of Kazakhstan.
Read more on Digitalbusiness.kz.
Share
