Artificial Intelligence is transforming the financial services industry, from algorithmic trading and risk management to fraud detection and customer personalization. Yet, adoption is constrained by regulatory requirements, explainability concerns, and the operational complexity of deploying AI in high-stakes environments. This tutorial provides a guide to building, evaluating, and deploying AI systems in financial services, with an emphasis on interpretable, compliant, and robust models. Participants will leave with a clear roadmap for developing AI solutions in financial services, practical tools for explainability, compliance, and deployment, and an awareness of common pitfalls in production AI systems and how to avoid them. Although not all compliance questions can be addressed, the tutorial will focus on certain regulatory insights into the EU AI Act. As the first comprehensive legal framework on AI, such insights may prove useful also for actors operating in other jurisdictions.
AI/ML researchers and practitioners interested in financial applications
Data scientists and engineers working in banks, fintechs, and regulatory technology
Financial analysts and risk managers seeking to understand AI capabilities
Compliance officers and legal teams exploring AI governance frameworks
Gabriele Mazzini is a pioneer and leading expert in Artificial Intelligence governance and regulation and a sought-after advisor, lecturer and public speaker across the world, with concurrent roles as a researcher at MIT Media Lab and fellow at MIT Connection Science initiative. Former Team Leader at the European Commission, he designed and led the drafting of the Commission EU AI Act and was the principal advisor during the legislative negotiations with the Parliament and the Council. He also shaped earlier policy work on the European approach to AI since 2017, including the White paper on the ecosystem of excellence and trust for AI and the work on liability for emerging technologies.
Previously, while based in New York he was a senior executive at the Millennium Villages Project, an initiative across sub-Saharan Africa pioneering science and technology-based interventions to alleviate extreme poverty, and helped advise start-ups on emergency communications and smart energy solutions. He also served in the European Parliament and the European Court of Justice. Gabriele holds law degrees from Harvard Law School, the University of Pavia and the Catholic University of Milan.
Svitlana Vyetrenko is a Founder and CEO of Outsampler, an AI startup that builds conversational agents for time series and tabular data. She is also Gutenberg Chair at the University of Strasbourg in France. Previously, Svitlana was an AI Research Director at JPMorgan Chase & Co., where she led a team developing cutting-edge AI products for financial services. Before that, she built client-facing AI-driven trading solutions at Goldman Sachs. She also served as an Adjunct Lecturer at Stanford University and a Lecturer at the University of California at Berkeley.
Svitlana holds a PhD in Applied and Computational Mathematics from California Institute of Technology, and has over 14 years of experience in the financial industry working on artificial intelligence and machine learning techniques for electronic trading. She previously co-organized workshops on ‘Machine Learning for Investor Modeling and Recommender Systems’ at ICAIF 2023, as well as ‘Simulation of Financial Markets and Economic Systems’ and ‘Foundation Models for Time Series: Exploring New Frontiers” at ICAIF 2024.
Tutorial Length:
TBD
Primary Contact:
Svitlana Vyetrenko svitlana@outsampler.com
By the end of this tutorial, participants will:
Understand the landscape of AI applications in financial services (trading, investment management, payments, loan operations).
Gain practical insights on certain regulatory compliance questions with a focus on the EU AI Act and the deployment and monitoring of production financial systems.
Learn the technical foundations of time series modeling, tabular data analysis, and natural language interfaces for financial data and explore model interpretability and explainability techniques tailored to regulated environments.
Part 1 – Setting the Stage: AI in Finance
Overview of AI applications in financial services: trading, credit scoring, fraud detection, AML, personalization
Part 2 – Regulatory & Compliance Landscape (lessons from the EU AI Act)
Why it is important to regulate use of AI in financial services
Risk-based approach and use cases in the financial sector
AI governance requirements under the EU AI Act for providers
AI governance requirements under the EU AI Act for deployers
Part 3 – Core Technical Foundations: Risks and Opportunities
Time Series Modeling: forecasting, anomaly detection, representation learning
Language-Data Integration: natural language querying and reporting from financial data
Large language models
Agentic workflows
Part 4 – Interpretability & Trust
Post-hoc explanation methods (e.g., SHAP, LIME, counterfactuals)
Intrinsically interpretable models for finance
Quantifying uncertainty and risk in model predictions
Part 5 – Deployment & Monitoring: Best Practices
Integrating AI into financial decision workflows
Continuous monitoring, drift detection, and retraining strategies
Case studies: scalable AI for fraud detection, explainable credit scoring, and market surveillance, etc
Part 6 – Hands-On / Interactive Session
Demo: building an explainable anomaly detection pipeline for financial transactions
ICAIF 2024: Tutorial “Evaluating and Rating AI Systems for Trust and Its Application to Finance”
ICAIF 2024: Workshop “Explainable AI in Finance”
Neurips 2024: Workshop on “Regulatable ML: Towards Bridging the Gaps between Machine Learning Research and Regulations”
ICML 2025: Workshop on “Technical AI Governance”