Selected projects

I have developed and improved products at different stages, from conceptualizing a machine learning product to optimizing the experience for millions of users and bringing products to new markets.


I have great vision and broad T-shape expertise due to my experience with various industry verticals.

  • EdTech application
    AI-powered reading tutor for children

  • Educational events
    Recommender system for social, educational events and MOOC courses
  • Logistics Company
    Price Forecasting and Margin Optimization
  • Automated UGC moderation
    Multi-label Text Classification

  • FinTech company
    Brought financial services to new markets
  • Media portal
    Consulting, Business Intelligence

How I work
"Success is no accident. It is hard work, perseverance, learning, studying, sacrifice and most of all, love of what you are doing or learning to do." - Pele
Initiation phase
  1. Identify project goals,
  2. Determine results,
  3. Understanding deliverables,
  4. Define risks and constraints,
  5. Estimate projects’ budgets
  6. Set priorities,
  7. Establish project scope (SOW),
  8. Submit project proposal (PR FAQ, BRD).
Planning phase
  1. Project roadmap
  2. Resources allocation,
  3. Milestones definition,
  4. Link product and data science metrics with business metrics,
  5. Detailed plan conduction,
  6. Solution architecture for the DS projects.
Execution phase
  1. Manage tasks,
  2. Track project progress,
  3. Review & adjust the project plan,
  4. Communicate project progress,
  5. Product testing.
Closure phase
  1. Handover of deliverables,
  2. Analyze performance,
  3. Release resources,
  4. Long-term monitoring.

The following are some of the products and projects I have been involved in over the past few years. My aim is to demonstrate the different product management and solution design approaches and techniques that my teams and I have employed, depending on the project's requirements. This includes the areas I specialize in:


Product management: Creating a product vision and strategy, developing detailed functional specifications, generating business hypotheses, and designing systems to ensure successful implementation. Recognizing the Minimum Viable Product (MVP) and product objectives. Identifying pain points and potential risks. Developing new features.


Product discovery: Applying design thinking to address user and business issues and enhance product/market fit.


Data Science Solution Architect: designing and developing data-driven solutions to complex business problems. Communicate the data science and machine learning solution to stakeholders and ensure that the solution meets the business needs.


Process establishing and optimising: Designing a development process that meets the requirements of both developers and other business stakeholders while also being tailored to the organization's culture and involving all relevant parties.


People management, mentoring, and growth: helping individuals and teams reach their full potential. Create and implement strategies to foster a positive work environment and promote employee engagement.

EdTech application
AI-powered reading tutor for children
Intro & Goal
The goal is to help children's enthusiasm for education so that they can create a better future.
Challenges
We had to take a concept and turn it into a web and mobile application that would assist kids in improving their reading skills through the use of voice recognition.
Solution
We have created a reading assistant. This platform automatically evaluates reading skills like fluency, dyslexia detection, current writing, understanding, and phonemic awareness.
My role
As a product manager, I was responsible for product vision, execution, stakeholder communication. I developed an idea into a product by formulating and validating business hypotheses and conducting CustDev. Through this process, during the business analysis phase, I identified a third customer - the parent.
I managed a team of 7 people, 4 of whom I had direct control over. This distributed, cross-functional team worked across multiple time zones to successfully complete the project on time and within budget.
As a NLP solution architect, I created an architecture that included ensembles of pre-processing, rule-based models, and speech-to-text models at both the phoneme and word levels.
Results
A fantastic product that is adopted by schools and raised investments.
Reduces teacher's manual assessment by about ten times.
Educational events
Recommender system for social, educational events and MOOC courses
Intro & Goal
A digital platform utilizing AI technologies links participants and optimizes their educational tracks.
Challenges
Develop a system recommending courses and events for new users about which there is practically no information. Establish metrics for evaluating the effectiveness of the system. Additionally, measure user engagement, such as click-through rate and user satisfaction.
Solution
A recommender system composed of an ensemble of models utilizing statistical methods, a Siamese neural network, and transformer-based language models was developed and integrated into the business process.
My role
I took the role of a project lead to coordinate the project's activities, including research, development, and delivery. My involvement in this project consists of four main tasks: structuring the workflow and documentation of the internal teams, creating a business plan, preparing a pitch deck, and conducting research.
I design architectures of 4 ML algorithms, 3 of which work in production. Implemented the hypothesis testing process within the ecosystem of products as a service. I have managed a cross-functional team of 6 people. And established the process of delivering the result to the customer.
Results
70% of clients followed the recommendations.
The Customer Satisfaction Index increased.
Increased completion rate of courses.
Logistics Company
Price Forecasting and Margin Optimization
Intro & Goal
A logistics transportation company specializing in finding the most cost-effective carrier for shipping orders.
Challenges
We aim to predict future prices based on a small amount of historical data. This task must address two main issues: negative margin and low conversion rate.
Solution
My team started by thoroughly studying business processes and then creating a model to predict the most advantageous price for both the customer and the carrier, which increases the chances of a successful transaction. This model is accessible as a web application and assists agents in closing profitable orders quickly. During the implementation, we discovered that a low data science metric does not always equate to a high overall profit, so I linked a data science metric with a business metric and adjusted our solution for maximum profitability.
My role
As a product manager, I was responsible for developing business and product strategies, budgets, roadmaps, discovery, execution & delivery, stakeholder management, and managing the team's health and performance by monitoring its members. Helped to manage project contracts and schedules.
I also created product requirements for the engineering and data science teams. Developed and presented reports on findings, improvements, statuses, and risk management. Managed and prioritizes the right backlogs while monitoring and improving team velocity. Tracked product use and impact on end users.
Results
The developed solution was incorporated into the business workflow, allowing up to 30% of all orders to be automated.
Automated UGC moderation system
Multi-label Text Classification for the media entertainment portal
Intro & Goal
A large information and entertainment portal where users publish their content wants to be sure that no prohibited data become public.
Challenges
The amount of content that should not be made public is growing quickly, faster than a human can keep up with. This puts a strain on moderators, who have to manage the increasing load.
Solution
Together with my team, we developed the classification system for texts. 18 different labels are detected, with an F1 Macro score of about 0.97 for the TOP 3 answers.
My team and I put a lot of effort into preparing, cleaning, and labeling data. For this, descriptions of categories and markup rules were prepared. We worked with an online crowdsourcing platform and external assessors for data labeling.
I linked a data science metric with a business metric to understand what metrics are critical to optimizing.
My role
As a product manager, I translated business needs into detailed feature requirements, data science hypotheses and drove programs from concept to implementation.
I optimize team performance, balance staff assignments, and improve processes. Here I got the unique experience of hiring and onboarding 4 new team members in just 2 weeks. So I helped to build and lead teams of problem-solvers of 8 cross-functional distributed specialists, 3 of them were part-time contractors.
In this project, I honed my skills in modern product planning, customer research, product discovery, and product development processes to effectively mentor and lead the team. I also developed and coached a highly competent team leader and data science team, emphasizing our principles and values in our work.
I created and executed a roadmap for the NLP program, which was fully delivered on time. I was also part of the responsible team for pre-sales and continued product marketing with potential and existing customers.
I formulated product specifications for the engineering and data science teams. I prepared and presented reports on discoveries, statuses, and risk management. I kept track of product use and its effect on the final customers.
Results
We completed the first working prototype within 2 weeks of initiating the project.
Reducing the time of moderation of one piece of information by more than 60%.
FinTech company
Brought financial services to new markets
Intro & Goal
A company offering financial services, including trading resources, to its customers.
Challenges
We had the strength to enter new markets. The challenge was to find new countries where we would expand. Due to legal regulations, I had one month to research and provided a report to CEO and Business owner.
Solution
I completed a comprehensive report this month which outlined the countries we should expand to and why. The report contained information from open sources such as estimated market volume, legal issues, competitors' market share based on advertisement and social activities, etc. I also developed a two-year strategic plan and a six-month plan of action. All of this was done by myself, relying solely on open sources. So I was ready to adapt as we implemented the program and received more information. Nine months later, the market leader released an industry report that had taken six specialists six months to compile. Surprisingly, about 80% of the conclusions of this report were in line with my findings, thus putting us nine months ahead of our competitors.
My role
As a part of the leadership team, I was responsible for revenue and product growth. Defined and drove the execution of the product strategy, ensuring a sustainable product-led business.
As a head of marketing and product, I provided comprehensive program management for the company. Prepared proposals and cost estimates for projects. I developed strategies, designed processes, and created product roadmaps. Additionally, I set financial objectives and managed the team. Additionally, I conducted user research. I successfully managed a cross-functional team of about 25 people, 5 of them directly.
Successfully prioritized product requirements and established achievable expectations regarding return on investment and timeline. I was analyzing the failures and successes of a product to set objectives for future development. I also worked to improve existing products based on customer feedback and requirements. Led collaborative meetings to create essential processes for developing new products. Evaluated product usability together with a UX researcher.
Results
I implemented a process to manage churn rate, retention, and unplanned retention expenses, resulting in a 30% reduction in the cost of acquiring a customer.
The company experienced a 12% increase in income over the course of 20 months by introducing the product to several Asian markets.
Optimization of employee reporting led to a 10% rise in employees' productivity.
A framework has also been developed for quickly and efficiently testing business hypotheses.
Media portal
Consulting, Business Intelligence
Intro & Goal
Entertainment portal (cinema, TV, music). The monthly audience (MAU) of the portal is 350,000 visitors.
Challenges
The company recruited me to gain insight into the aims and targets of each department, create data-driven business dashboards, and also aid in formulating a marketing plan.
Solution
I analyzed a company’s business operations, identified problems, and recommended solutions. I also help organizations to develop marketing strategies, implement changes, and evaluate the effectiveness of those changes.
My role
As a consultant, I facilitated meetings and workshops to create processes for prioritizing and measuring new hypotheses.
I suggested a product development strategy and broke it into a tactical plan and epics, which was approved by the CEO. So for the next step, I monitored the execution of the created strategy.
I was in charge of designing and defending the application's monetization schemes to shareholders.
I developed and deployed analytical and business dashboards to meet the needs of various departments, including marketing, sales, technical specialists, and management, as well as providing training to personnel on the utilization of new tools.
My task was to identify issues based on the data from analytics and suggest feasible solutions that would take into consideration the user interface of the web, iOS, and Android platforms.
Results
I created a framework for fast and efficient verifying business hypotheses, leading to the doubling of average view per user (AVU). (AVU in media is like an average check in e-commerce).
The mobile app was ranked number one in the overall AppStore rating and the top 50 of Google Play.