Balancing Quality, Profitability, and AI in Online Learning: Insights from a Chief Methodology Officer and Data Science Expert Ilnar Shafigullin, as the head of the franchise creation department at a leading company in this market, established a new industry standard with the introduction of the Franchise Management Guide

Ilnar Shafigullin has over 10 years of experience in education. He graduated from the Mechanics, Mathematics, and Economics Faculties at Kazan State University, a prestigious institution renowned across the CIS for its rich history and academic heritage. After graduating, he pursued further academic endeavors and entered graduate school, where he earned a Candidate of Science degree in Physics and Mathematics (equal to an American Ph.D.), marking the start of his teaching career.

In the following years, he lectured and led seminars for students in subjects such as mathematical analysis, optimization methods, algorithms and data structures, and computer science. He also engaged in advanced scientific research, participated in numerous international conferences, and published over 10 articles in peer-reviewed journals.

Beyond academia, Ilnar's teaching influence extended to the corporate sector. As the head of the franchise creation department at a leading company in this market, he established a new industry standard with the introduction of the "Franchise Management Guide." This innovative guide provided entrepreneurs with comprehensive insights encompassing every aspect of starting a business—from client acquisition to effective marketing campaigns.

"Even in this field, I continued to teach, as the essence of my work remained the same—explaining what I had learned to another person who had not yet mastered this knowledge," Ilnar states.

Simultaneously, Ilnar never ceased learning. He became an expert in Data Science and served as a Machine Learning engineer at one of the largest banks in the Republic of Tatarstan. There, he demonstrated his skills in AI and computer vision, playing a key role in implementing a project aimed at tracking passengers on public transport. He utilized Python, PyTorch, and other technologies to perform various tasks—from model selection and data collection to deploying systems on Nvidia Jetson Nano and other platforms.

Transforming Online Education at GeekBrains

By the time Ilnar joined GeekBrains, the education market included a variety of players: companies and individual specialists offered online, offline, and hybrid learning in synchronous or asynchronous formats, individually or in groups.

GeekBrains specializes in large-scale synchronous online IT education and long-term programs (usually over a year) designed for tens of thousands of students. These programs were conducted in real-time, where students not only watched pre-recorded lectures and seminars and completed homework with automatic or manual checks, but also participated in classes with peers, solving problems under the guidance of experienced instructors.

This approach ensured greater engagement and higher completion rates but was harder to scale. For a company teaching 50 students, one expert would suffice; for 5,000 students, 100 experienced instructors were needed, which was extremely challenging to find. Additionally, operational costs would multiply.

Ilnar faced a huge challenge: to maintain the synchronous learning format while improving and scaling the programs. Working at the company became an opportunity to apply his extensive experience to transform the additional online education market. He analyzed GeekBrains' current products, identified improvement points, and, together with his team, developed and implemented an innovative product—the flexible "Developer" program with a tree structure and a combination of synchronous and asynchronous learning elements, later adopted by many competitors.

Education Tailored to Client Goals

Prior to the COVID-19 pandemic, all programs within the industry followed a singular principle of linear learning along unrelated paths. Students were required to choose a technology and specialization from the beginning, whether it was Python development, SQL data analysis, or DevOps.

This approach placed adult professionals, typically looking to change careers, in the position of high school graduates who often choose college paths based on parental advice, peer experiences, and teacher recommendations, without a clear understanding of their own aspirations and career prospects. Unsurprisingly, about 60% of college students would change their major if given the chance to start anew. Similar regrets are prevalent among clients of additional online education programs.

In the "Developer" program, Ilnar applied a different approach. The first two months of training followed a standardized curriculum for all students. Students were introduced to the basics of programming, version control, languages, databases, and practical tasks. Only after completion did they choose one of five specializations: programmer, analyst, tester, project manager, or product manager. After a few more months, students could choose a technical specialization from dozens of available options, such as system administrator or a Fullstack developer.

Thus, training began with a fundamental program and ended with the acquisition of specialized skills. This approach also ensured the relevance of students' knowledge upon entering the job market.

Consistent Launches and Sustainable Business Growth

The new format benefited both students and the company. The "Developer" program simplified the preparation of educational materials by eliminating the need to package similar content for separate paths.

It also solved the problem of group filling. When a company focused on synchronous learning creates dozens of diverse courses, it may encounter insufficient student enrollment for profitable launches across the board, due to uneven demand for different technologies. If, for instance, only 5 enroll out of the expected 20, a decision must be made: wait for additional enrollments or refund those who have already paid. Both options can result in dissatisfaction, potentially leading clients to seek out competitors.

The "Developer" program helped retain students with specific, niche specializations. It provided a unified starting point, and new groups began to launch biweekly. To streamline the search process of experts, Ilnar implemented another innovation within the program: asynchronous lectures recorded in a professional studio. Meanwhile, seminars continued synchronously, allowing students to solve problems and ask questions to instructors who, in turn, provided explanations using pre-prepared materials. This approach effectively minimized the need for an extensive expert search.

Balancing Quality and Profitability

Within the project, Ilnar not only identified problems, but also developed the IT competency matrix and methodology, designed trajectories, and analyzed the effectiveness of educational programs.

Ilnar notes that analysis should evaluate both the methodological and business aspects. It's important to consider parameters such as course completion rates, class attendance, and homework submission rates. Additionally, intermediate assessments should be conducted to gauge material comprehension.

"On the other hand, economic aspects should be taken into consideration, with a focus on group filling and training profitability. Although small synchronous learning groups ensure maximum attendance, this option won't bring profit to the company. It's necessary to strike a balance between quality and positive unit economics."

Ilnar was also responsible for organizing the effective work of methodologists and instructors. He highlights that his main principle is to learn first, then teach others. "At the program's start, I recorded and conducted lectures and seminars on programming basics for groups ranging from a few dozen to several hundred people," states the expert.

Gradually, he expanded the team by inviting new methodologists and instructors. Each new expert underwent personal training by Ilnar—he reviewed lecture content, made corrections, and conveyed his vision. Only after mastering the concept did the employee begin to work independently. According to Ilnar, this approach to building a team can be called self-duplication. It is the most effective way to scale an innovative idea without losing its essence.

AI-Enhanced Learning at Every Stage of Education

In discussing trends in the EdTech market, Ilnar emphasizes the crucial role of neural networks, including generative AI. He is well-versed in this field, having created an MVP AI product for English language learning. Within this product, neural networks generated user profiles and provided tasks aimed at improving weak areas. Correct solutions were rewarded with points, while mistakes deducted them.

This dynamic system was later adopted by a state digital educational platform. Ilnar notes that the concept can be applied to any field, such as programming, analytics, or testing. "The most challenging aspect," says Ilnar, "is preparing the competency framework necessary for effective training."

The expert believes that neural networks can be applied at any stage of the educational process. For instance, AI can assist with career guidance by analyzing tests and even verbal responses to questions, comparing the information with market trends, and suggesting whether an individual should become a programmer or project manager. AI can also identify the best training service for users based on their schedule and format preferences.

During the learning process, AI can assist lecturers by accessing lecture materials to answer student questions in line with the teaching methodology. Similarly, an AI assistant can facilitate practice sessions, helping students independently reach correct solutions, thereby saving instructors' time and reducing companies' overall costs.

Products like these are gradually appearing in the market. Google recently introduced an update to its Circle to Search function, now featuring a specialized tool to help students with math and physics homework. The function visualizes a step-by-step process rather than providing immediate answers.

Another interesting use of AI is task generation. If a student struggles with a particular type of task, they can ask the AI to generate additional similar tasks. Moreover, these tasks can be highly personalized. This capability is especially valuable for adult learners in additional education programs, who often have diverse backgrounds—the same material may be studied by both an accountant and a taxi driver. In such cases, AI can offer the first student a task related to financial calculations and the second student a task related to fuel consumption.

In the future, AI will be able to continuously update a user's profile and dynamically adapt their learning program. Upon completion, the user will receive not only a diploma and a resume featuring a general list of skills, but also a detailed analysis of competencies across dozens of parameters. This approach will significantly improve the matching of candidates with future employers.

Ilnar emphasizes that AI will not only revolutionize the approach to online education but also reshape content of educational programs. With technological advancement, new professions and individual skills will emerge that specialists need to master—such as creating product cards on marketplaces using AI text and image generators. To stay relevant, online educational companies must offer their clients training in these new areas.