Michael Dubovoy

Built and owned sales and demand forecasting across ~50,000 time series using deep-learning models (PyTorch, Darts). Automated the end-to-end pipeline on Google Cloud Vertex AI, replacing manual processes.

Built backend data pipelines and services for an enterprise AI chatbot, worked with ontologies and SPARQL. Conducted R&D on airline ticket-price prediction and transfer learning for DB column classification.

Built an ML pipeline for Bayer AG's agricultural research, including statistical tests and hypothesis checks. Developed distributed pre-processing pipelines and visualizations for HERE Technologies demos.

Developed customer-scoring ML models to reduce loss ratio for Trov, an on-demand insurance client. Built and tuned an NLP topic model (LDA) on Wikipedia at scale using Apache Spark and AWS EMR.

Conducted R&D of a real-time dynamic pricing model based on demand behavior and built a behavioral testing simulator.

Developed ATM cash-demand forecasting to cut costs and hold downtime KPIs. Created business forecasting models and automated reporting.

Grade: Distinction

Activities and societies: King's College London Graduate Scholarship (1 of ~80).

MSc in Financial Mathematics with Distinction. Dissertation on 'Risk Measures and Valuation'.

Grade: Distinction

BSc (Hons) in Mathematics with Distinction.