Dr. Vladimir Vutov

- Developed binary classification pipelines utilising XGBoost and CatBoost, improving model performance via Optuna-based hyperparameter optimisation - Contributed to a transformer-based classification framework, culminating in a conference paper published in the field of AI - Created a hybrid RAG pipeline for scientific paper Q&A, integrating advanced retrieval techniques and LLM-led answer validation

- Developed large-scale multiple testing procedures for feature selection in high-dimensional data, resulting in two first-author publications - Co-developed a framework applying topological data analysis (TDA) to denoise and compress high-dimensional mass spectrometry imaging (MSI) data for automated cancer classification; published in a peer-reviewed journal - Contributed to optimising robotic grasp position inference by developing a probabilistic ML framework in partnership with the Institute for AI

- Engineered a custom recommender system to help users find their ideal real estate properties - Designed and implemented prediction frameworks to generate real estate recommendations in production using Python

- Leveraged machine learning methods to extract patterns from client datasets and address specific business questions - Applied predictive modelling techniques (e.g., GLMs, random forests, quantile regression) to improve marketing campaign performance - Created dashboards, visualisations, and reports to communicate analytical results