Anton Olenin

Initially, it was a huge monolith architecture with many state managers. It used pure JavaScript with no type definitions and was maintained by many teams, sometimes located in different countries and time zones. All of this was bundled with an old webpack with no code splitting. Each release took one day in case everything was ok, and the releases blocked any other releases of the other teams.

After splitting into a micro-frontend architecture, the monolith application became the core of our single-spa application. Each team could gradually extract relevant pages to separate repositories, connect to the core application, and choose what technologies to use. The only peer dependency was React, and finally, most of the teams introduced TypeScript. Each team also had its cozy application, not larger than 100 kilobytes and within 4-5 pages, independent releases, and even more independent release processes. Lastly, each team could add its own steps to CI/CD pipelines, such as screenshot tests or alerts in case of sharply increased bundle sizes after specific commits to the main branch.

When I took over the project, the search worked only on one website page. The search results were rendered on the server side and returned to the client as HTML. The client-side parsed HTML to JSON and then rendered.

After working on the project, the search bar became available on all pages with autocomplete, search facets, and an extended search functionality having over 20 filters with complex inputs. All data was provided within 1-2 network calls and worked at least twice faster. All search-related metrics grew since customers got a better user experience and could perform more search requests.

The task was to create from scratch a prototype of a "local YouTube-type platform" for Kazakhstan to get a budget for the startup inside a big tech company. There were two of us working on this project, and I handled the front end. After one month of work, we demonstrated our minimum viable product (MVP), which included signing up and logging in by email, searching videos by keywords, a listing of videos with infinite scroll, a detailed video page, the like and comment functionalities, and a simple recommendation system based on likes. The project received the needed resources and was considered one of the company's most successful products. It is still being developed up until now.