Keven F.

– Contributing to a cutting-edge document extraction system using LLM-based approaches, facilitating the processing of over 80K documents per week. – Developed an AI-powered collection assistant that increased operational efficiency by 20%, winning FactSet's global engineering hackathon and leading to its deployment. – Led the design and implementation of an end-to-end ML pipeline on AWS, supporting 20 ML engineers with scalable training and serving processes. – Trained, evaluated, and deployed small LLMs, boosting document extraction and classification performance by 20% and reducing inference costs by 40% for fund and company documents. – Key developer of a scalable, rapid entity-linking service (FactSet Concordance), mapping external data to 20M+ FactSet entities, improving data integrity and accessibility.

– Implemented advanced NLP techniques to match and classify internal and external data referential (using Nltk, Gensim, Sklearn and Spacy). – Built robust data pipelines and analytics reports to automate Client Account Plans and Strategic Business Reviews, thereby boosting efficiency for over 20 bankers in global banking coverage teams.

– Defined KPIs to support heads of Portfolio Control teams in their management activities. – Created dashboards to track the evolution of NAV, AUM, and risk indicators of portfolios.