Oleh – Python, Django, SQL

RAG Chatbot for digital agency website that provides all access to the knowledge base that is built on the website's content and some additional documents provided by the agency

• Solely developed a RAG Chatbot for a digital agency's website, serving as the central developer responsible for the entire backend construction of the application
• Employed FastAPI to build robust, efficient, and high-performing backend services, ensuring seamless interactions and prompt responses from the chatbot
• Leveraged Qdrant to create a powerful vector search engine that significantly enhances the chatbot's ability to retrieve accurate and relevant information
• Deployed and managed the chatbot services on AWS, ensuring reliable uptime and high availability to meet the agency's operational needs
• Integrated Langchain; Using of the OpenAI to supplement the chatbot's functionality, facilitating advanced AI and machine learning capabilities and enriching the user experience
• Achieved seamless extraction and utilization of knowledge from the digital agency's content and documents, enabling the chatbot to deliver precise answers, driving user engagement and satisfaction

The idea was to reduce human efforts on the initial candidate screening. Candidates were using our AI-based application to pass the initial screening

• Researched different approaches for extracting information from raw inputs
• Wrote integration tests
• Integrated logs sharing to the CloudWatch
• Performed bugfixes

This was s platform for creating, assigning, delegating, and managing tasks in a specific domain. Tasks were moving through task flow (different statuses, etc).

• Designed the core architecture for TMS
• Created several dataflow diagrams
• Implemented new functionality
• Fixed bugs, CI-CD​

Platform used for textual processing of recorded calls.

It was implemented as integration for a bigger product where data comes from.

• Customized Django admin
• Worked on the flow of assigned pieces of dialogs to users for execution and for subsequent review
• Worked on the flow of calculation of payments for text editors and reviewers "Counted the number of changes in the text" made by the editor
• Implemented bug report functionality for editors and reviewers
• Integrated with third-party services

This marketplace was a set of web applications that integrated 3rd party e-commerce services and platforms and provided special functionality.

• Integrated with 3rd party services (Storm)
• Implemented new features
• Wrote integration tests
• Performed troubleshooting, refactoring, fixed bugs

Several web applications for clients that need to handle different types of legal operations, such as invoices, relationships with vendors, etc.

• Created extended tests as a one-step to upgrade Python versions
• Developed new features
• Performed troubleshooting, refactoring, fixed bugs

The Core idea of the project was to help food retailers to plan the positioning of the products inside a shop space. Collected data then were used to model a customer's behaviors and to build some analytics. In general, the process of digitization of a product position was handled almost manually and it takes a lot of human time. The idea was to automate this workflow using AI and Transfer learning and DCNN

Oleh developed a chatbot that helps to collect some data for training and testing models. In general, this chatbot inspected an image for a barcode (of the specific product) and recorded a video where the product is rotating, using different focus, zoom and shooting angle, etc. Using the barcode of a product team defined class labels for all data points that were extracted from the recorded video.

Oleh also trained DCNN models of popular architectures like ImageNet and ResNet using the transfer learning approach; played with fine-tuning.

Web application for students, parents, and doctors. This platform was used for screening and controlling the student's health, to detect early stages of diseases, and prevent them.

• Integrated the main platform using OAuth2.0
• Performed troubleshooting
• Refactored and implemented new features, fixed bugs

Web application for art and antique dealers.

• Crawled and prepared data
• Implemented Flask app to manage jobs related to crawling
• Used Crawlera for concurrent requests and proxies rotations
• Reversed engineering for crawling directly API of needed sources
• Added Auto Re-Captcha solving
• Performed troubleshooting, refactoring, fixed bugs