Guess word

A game where users guess the word of the day or week by relying on the semantic similarity percentage between words. The system employs text processing algorithms for accurate similarity detection, supports social media registration, tracks user statistics, and features a leaderboard for competitive engagement.

Interactive 3D Tours and Construction Analytics

We developed a platform for creating 3D tours of residential complexes and private houses using ML technology. The project combines 360-degree photography and machine learning to create virtual tours and deliver comprehensive analytics.

I was responsible for the complete backend development, including architecture planning and implementation of all backend components: APIs, data processing, orchestration with Airflow, Docker containerization, and the deployment and maintenance of ML models.

Browser extensions with AI

• Text summarizer;
• Image to text;
• Image to Prompt

My project contribution involved creating a flexible microservice architecture for browser extensions, ensuring high performance, reliability, and scalability.

Moreover, I was responsible for selecting and analyzing project implementation tools, serving ML models, and applications containerization.

Services for applying machine learning models to work with images.

We have created a platform that allows for the use of machine learning for automatic image processing and analysis. Our service can remove backgrounds from images, automatically generate descriptions, and extract tags, making it easy and efficient to work with large volumes of photos.

My role:

• Participation from the very beginning, including design, technology selection, and development;
• Models and applications containerization;
• API & Telegram Bot development;
• Testing and documentation;
• Payment system integration.

Bot for maintaining a healthy diet

As part of the project to develop a bot, I deployed the llama-2-13b model and implemented an API for analyzing users' meal photos.

The bot analyzes images of dishes, providing information about the composition, calorie content, and nutrient content, and also generates a daily report in PDF format.

In addition to implementing the functionality, I also integrated the YooKassa payment service and participated in developing an algorithm for automatically generating a meal plan based on user parameters and goals.

Bot for Nutrition Center managers

A bot for automatic generation of weekly menu according to nutrition protocols.

My role:

• Bot automatically generates personalized weekly menu for clients based on their nutrition protocols and allergies.
• Used the Yandex GPT API to create text descriptions of dishes and integrated with Kandinsky to provide photos for each dish.

Automation of data collection and analysis for products in construction stores

This project was aimed at automated gathering of information about products in construction stores for the purpose of price analysis. Our team developed a system for parsing data from websites of construction stores and storing them in Excel and PostgreSQL formats.

My role:

• Parsers development using API, FTP libraries, Selenium and BS4;
• Data processing in JSON and XML formats and database storing;
• Parsers execution automation, data storage and reports delivery using Apache Airflow.

Construction monitoring

The Smart Monitoring project is an artificial intelligence-based tool for real-time construction monitoring and safety compliance. The tool allows for the analysis of video streams from cameras, identification of construction stages and reporting the presence of protective barriers.

My role:

• Admin panel for results visualization and models operations maintaining;
• Automated report delivery through Telegram;
• Bot development for uploading images from drones and special equipment;
• Integration with customer systems.

RealEstate Data Aggregator

The project aims at creation of service that automatically collects information about apartments available for sale.

My role:

• Development and configuration of parsers for collecting data about apartments;
• Setting up a parser launch schedule;
• Organization of data recording in Airtable and PostgreSQL databases;
• Notifications configuration and sending in Telegram;
• Project deploy on Google Cloud platform.

Decor monitoring

An artificial intelligence-based tool for automated monitoring of interior finishing in construction.

My role:

• Containerization of models and project components to ensure convenient deployment and scalability;
• Development of visualization and access to monitoring results in the corporate service;
• Implementation of a notification system and reporting mechanism;
• Integration with customer systems.

Aggregator platform for NFT loans.

The Lendz project is an equivalent of 'Skyscanner' for NFT lending. Lendz helps borrowers secure cryptocurrency loans against NFT assets on the most favorable terms through offer aggregation.

My role:

• Development of a microservice for collecting data about loans from NFT platforms;
• Processing and aggregation of loan data;
• Implementation of a notification system and reporting mechanism;
• Development and creation of a Telegram bot for interacting with the system;
• Personalization of notification settings for users.

Service for searching for actors' faces with additional image tagging.

The project aims to enhance user experience by implementing a feature to search for actors and movies based on photographs. This allows users to easily find information about desired actors and movies they have starred in.

My role:

• Image and video acquisition for model training;
• Creation of a bypass service for gathering links from Telegram channels and downloading data from cloud storage.

Assessment of OSAGO (Motor Third-Party Liability Insurance) and fraud analysis using geodata.

The project is a service for evaluating the cost of OSAGO policies, predicting customer influx and outflow, and analyzing insurance events to detect fraud.

My role:

• Creation of an external health service to ensure the continuous operation of the main service, including automatic restart and alerting the team about failures;
• Development of infrastructure for collecting key metrics of models, their visualization, and sending daily reports;
• Ensuring data storage and validation.

Corporate Information System for company employees.

• Application of knowledge obtained from Java and Spring courses for service development;
• Implementation of functionality for controlling access to confidential employee data and projects using LDAP protocol.

Sales aggregation platform

The project is a platform that brings together fashion brands and special offers from online stores, providing a personalized shopping experience.

My role:

• Writing tests to familiarize with the project's structure and learn its functionality;
• Learning Python and Django REST framework;
• Application of acquired knowledge in practice, including currency support and conversions, proxy usage, and integration with Yandex XML feeds.