Research and Implementation of Yandex.Eat's Cart Functionality in the Infrastructure of Yandex.Market

This thesis is dedicated to examining and implementing the integration of Eda and Lavka functionalities into the infrastructure of the Yandex.Market. Historically, all user query handling logic and infrastructure at Yandex.Market has fallen on the client side, leading to significant delays in page loading, navigation, product card views, cart checks, and related functionalities. The goal of this work is to develop connected retail (CR) at the Market through integrating the functionalities of Eda and Lavka. This dissertation considers approaches to implementing a Back-End Driven UI (BDUI) architecture opposing FDUI complexity, where the main UI control logic is managed server-side. Previously, the orchestration of cart queries and order placement, including mobile applications and browsers on various platforms, rested on the client. This has led to several issues, including the multiplication of code across different clients, delayed time-to-market (TTM) for releasing new functionalities, and degraded user experiences due to noticeable page loading delays (with the 95th percentile of page load times reaching 5 seconds). The project focuses on integrating the cart into new services - Aktualizator and an optimizer for CR which corresponds to BDUI. The main task is part of increasing the conversion coefficient from 11% to 23% of Lavka carts in Yandex.Market. It also plans to enhance metrics for recommendations and Market orders. This project aims to implement a parallel processing flow of the cart, which could handle requests for different interoperability scenarios with various services