Roman Shevchenko

Helping companies transform complex workflows into AI-powered systems through production-grade LLM, RAG and Multi-Agent architectures. Hands-on development of reliable AI solutions across Healthcare, Life Sciences and enterprise environments — from architecture design to production deployment. Key achievements: - Built an AI platform for veterinary clinics, combining X-ray analysis (5,000+ medical images), multimodal AI models and automated radiology report generation. - Developed a real-time ASR system for clinical documentation, reducing doctor consultation time by 50%. - Designed and deployed production RAG pipelines for pharma use cases, improving retrieval accuracy by 40% and reducing inference costs by 45%. - Designed multi-agent architectures to automate complex veterinary workflows using specialised AI agents.

Building and deploying production AI systems for Life Sciences and Pharma organisations, helping teams automate knowledge-intensive workflows and improve operational efficiency. Key achievements: Architected and deployed LLM-powered knowledge platforms using RAG and multi-agent architectures for Pharma and Healthcare use cases. Built AI assistants and intelligent workflows to support pharmaceutical teams, automate information retrieval and improve internal knowledge access. Designed AI automation solutions for document processing and information extraction, reducing manual effort by 40%. Integrated AI capabilities into existing business workflows, accelerating content creation and improving operational efficiency. Led AI engineering initiatives, providing technical direction, architecture decisions and mentorship for AI development teams. Designed scalable production AI infrastructure, optimising deployment reliability, performance and inference costs using modern cloud-native technologies. Tech stack: Python, LangChain, LangGraph, LlamaIndex, Ollama, Hugging Face, RAG, Multi-Agent Systems, Docker, Kubernetes, AWS, CI/CD.

Designed and deployed production machine learning solutions for banking and enterprise environments, covering the full ML lifecycle — from business problem formulation and data analysis to model deployment and integration into operational systems. Key achievements: Built predictive analytics platforms for customer behaviour modelling and integrated ML solutions into banking CRM systems to support data-driven decision-making. Developed an AI-powered resume matching system for corporate HR platforms, improving candidate selection efficiency through NLP-based matching algorithms. Designed predictive HR analytics solutions to identify employee burnout and attrition risks, supporting workforce management strategies. Created a geospatial analytics solution to optimise ATM placement, generating quarterly savings of over 1 billion rubles. Automated ML workflows and developed scalable data pipelines for production model deployment. Tech stack: Python, PyTorch, Scikit-learn, CatBoost, Pandas, SQL, Spark, Hadoop, Teradata, Kubernetes, Docker, CI/CD.

Leading Data Science team (6 DS and 2 DE). Construction of quantitative models for predicting price movements for stocks and derivatives. Creating an internal product for making trading decisions based on machine learning algorithms using fundamental and technical data about assets. Development and implementation of a risk management system. Accomplishments: - Conceived, developed, and implemented a predictive analytics system for analysing and forecasting stock price movements for over 15,000 global companies, resulting in a 30% annual profit increase for the fund.

Development and implementation of price forecasting models for various asset classes (cryptocurrencies, commodities, stocks, derivatives). Building predictive analytics systems based on machine learning (classification, regression, anomaly search) for trading operations. Formation and verification of hypotheses. Analysis of the effectiveness of CRM campaigns. Conducting analytical research for working with clients, marketing and financial management.

Support for investment decision-making in terms of creating and updating internal databases of instruments available for investment. Creation of predictive financial models of companies – issuers of stocks and bonds (profit/loss, balance of assets/liabilities, cash flow). Development and support of fair value measurement models for various financial instruments, the formation of analytical notes and professional judgments based on the results of such an assessment.