- Developed and Deployed an Intelligent Chatbot: Created a chatbot using Flask, later migrating to Socket.IO for real-time communication, enhancing user engagement and support.- Implemented Semantic Search with Vector Embeddings: Utilized vector databases (e.g., Chroma) to store embeddings, enabling efficient semantic retrieval of information and improving response relevance.- Advanced Prompt Engineering: Fine-tuned language models and crafted prompts to optimize the chatbot's conversational abilities and accuracy in understanding user queries.- Natural Language Processing Optimization: Developed NLP pipelines for text preprocessing, including tokenization and language detection, to handle multilingual inputs effectively.- Production Deployment and Scalability: Deployed the chatbot in a scalable environment, ensuring high performance and the ability to handle increased user traffic.

Thesis: Automatic Segmentation of Asbestos Fibers in Low-Magnification Electron Microscopy Images.– Developed a U-net-based CNN algorithm for the automatic segmentation of asbestos fibers in electron microscopy images.– Addressed limited annotated data through data augmentation techniques like cropping and rotations, improving model training.– Optimized performance using a combination of Binary Cross-Entropy and Dice loss, and employed OneCycle learning rate schedulers.– Achieved an ROC AUC of 0.9864, demonstrating the effectiveness of the model for asbestos detection.

This company is a leading expert in the IT consulting market and the development and implementation of effective solutions for business automation. Big Data Analysis and Reporting:– Conducted real-time big data analysis using PySpark to process streaming data from client websites, focusing on user activity and purchase behavior.– Developed AIRFLOW scripts for nightly data uploads, ensuring consistent data availability.– Generated monthly trend reports and vital statistics for clients using PowerBI, highlighting key performance indicators and actionable insights.Client Communication and Data-Driven Strategy:– Collaborated closely with clients and team members to identify project objectives and essential predictive data requirements.– Regularly collected, analyzed, and presented data findings using PySpark and Pandas, maintaining transparent communication about progress and potential challenges.– Explored and tested various hypotheses based on the data using Machine Learning approaches, contributing to the development of informed, data-driven strategies

BlackVR Is an IT company for the development of AR and VR technology, 360 video and websites production. My main tasks were - data analysis with excel and R, - leading projects and the development team,- helping to organize and conduct events involving with virtual reality helmets use and managing with full logistics for such events.