Nick Kobets

Create and deploy a fully functional Retrieval-Augmented Generation (RAG) system using advanced Large Language Models and embedding models, enabling accurate retrieval and generation of contextually relevant responses. Optimize and maintain a vector database for text embeddings, enhance the efficiency of semantic search and retrieval operations. Improve model performance for domain-specific queries through fine-tuning, training embedding models and prompt engineering to optimize query responses. Integrate RAG systems into existing application, enhancing user experience and streamlining product workflows.

Develop AI agents using LangChain and LangGraph for automated information extraction and business process optimization, leveraging prompt engineering for precise and context-aware task execution. Integrate AI systems into existing workflows using tools like LangChain, REST API, Docker, vector databases and AWS services for seamless automation and scalability.

Contributed to developing chatbot for Canada immigration process using OpenAI models and Rasa framework. Key responsibilities: mapping user's journey, web scraping, data preprocessing, testing and deployment. Collaborated on developing a website to analyze cough severity through audio records using ML. Key responsibilities: data preprocessing, audio preparation & feature extractions, exploring various neural network architectures suitable for audio analysis tasks, testing and deployment. Participated in creating mobile app targeting illegal deforestation, utilizing ML frameworks and real-time object detection YOLO model. Key responsibilities: data collection, preprocessing and analysis, model experimentation, model fine-tuning and optimization, evaluation, testing and deployment. Team created PoC AI tool to create photorealistic digital likenesses of individual by training Stable Diffusion model, using Deepfake technologies for accurate and ethical replication. Worked with sequential data utilising neural networks (RNN, LSTM, GRU models), enhanced skills in handling time-series data.