The half day workshop on Wednesday 30th September is complimentary to all delegates: 13.45 – 17.15
Course Overview
AI model risk is a new discipline, where regulatory requirements and best practices borrow heavily from other types of model risk. The practices adopted from traditional model risk management (MRM) are not always capable of dealing with the unique characteristics and challenges of AI model risk. Effective measurement, reporting and mitigation of AI model risk requires combining novel, AI-specific risk metrics and techniques with traditional risk management practices.
In the first part of the course, Alexander Sokol will present practical and effective techniques for the quantitative measurement and reporting of AI model risk using both well-established and novel metrics. In the second part, Alexander will leverage his award winning research on behavioural psychology of AI to describe practical and effective techniques for mitigating AI model risk and increasing reliability of AI-based workflows.
- Implement AI-based workflows based on use cases of practical
importance to banking and asset management - Measure aleatoric and epistemic risk of these workflows
- Mitigate the sources of both types of risk
- Measure the resulting improvements in model risk and reliability metrics
Learning Outcomes
- Understand quantitative measurement and reporting of AI model risk
- Learn the key metrics for the accuracy and reliability of AI
- Learn how to mitigate AI model risk using advanced techniques
from behavioural psychology - Learn to design reliable AI-based workflows
Session 1: 13.45 – 15.15
Coffee Break: 15.15 – 15.30
Session 2: 15.30 – 17.00
Q&A: 17.00 – 17.15
Alexander Sokol:
Head of Quant Research, CompatibL
Alexander Sokol:
Alexander Sokol: Head of Quant Research, CompatibL
Alexander Sokol is the founder, Executive Chairman, and Head of Quant Research at CompatibL, a trading and risk technology company. He is also a co-founder of Numerix, where he served as CTO from 1996 to 2003.
Alexander won the 2018 Quant of the Year Award together with Leif Andersen and Michael Pykhtin for their joint work revealing the true scale of the settlement gap risk that remains in the presence of initial margin. Alexander’s other notable research contributions include systemic wrong-way risk (with Michael Pykhtin), joint measure models and the local price of risk (with John Hull and Alan White), the use of autoencoder manifolds for interest rate modelling (with Andrei Lyashenko and Fabio Mercurio), and the mean reversion skew.
Alexander graduated from high school at the age of 14 and earned a PhD from the L.D. Landau Institute for Theoretical Physics at the age of 22. He was the winner of the USSR Academy of Sciences Medal for Best Student Research of the Year in 1988.
Course Overview
AI adoption is no longer the central question for financial institutions. Most firms are already experimenting with generative AI, copilots, coding assistants, and increasingly agentic systems. The more important question is whether these systems are changing core business workflows, or whether they are merely adding another interface on top of existing processes.
This half day course examines how agentic AI can move from productivity overlay to durable infrastructure in finance. The course focuses on the design of agentic systems as “digital organizations”: structured systems in which specialized agents reason across subtasks, use tools, coordinate decisions, and operate within defined control boundaries.
Participants will explore which financial workflows are suitable for agentic automation, how task structure influences topology choice, and why single agent, supervisor, hierarchical, independent, and hybrid architectures behave differently in production. The course then moves from design to operational reality, covering the coordination tax, retries, latency, validation gates, correction loops, threat modeling, and production controls.
The course provides a practical architecture lens for financial institutions that want to move beyond AI experimentation and toward reliable, governable, and economically sustainable agentic adoption.
Part 1: Why agent infrastructure matters: 13.45 to 14.15
Adoption versus impact, horizontal copilots versus vertical workflow transformation.
Part 2: Finance workflows and task decomposition: 14.15 to 14.40
Which finance tasks are agent suitable and why.
Part 3: Topology choice: 14.40 to 15.15
Single agent, supervisor, hierarchical, independent, hybrid, mapped to finance use cases.
Part 4: Reliability in production: 15.30 to 16.00
Coordination tax, retries, latency, validation gates, correction loops.
Part 5: Threat modeling for agent systems: 16.00 to 16.30
Layered threat model, CIA mapping, model to orchestration to tool to memory propagation.
Part 6: Production controls: 16.30 to 17.00
Guardrails, red teaming, memory controls, HITL, observability, governance.
Session 1: 13.45 – 15.15
Coffee Break: 15.15 – 15.30
Session 2: 15.30 – 17.00
Q&A: 17.00 – 17.15
Nicole Königstein:
Chief Data Scientist, Head of AI & Quant Research, Wyden Capital AG
Nicole Königstein:
Nicole Königstein: Chief Data Scientist, Head of AI & Quant Research, Wyden Capital AG
Nicole Königstein is a distinguished Data Scientist and Quantitative Researcher, currently working as Data Science and Technology Lead at impactvise, an ESG analytics company, and as Head of AI and Quantitative Research at Quantmate, an innovative FinTech startup focused on alternative data in predictive modeling. Alongside her roles in these organizations, she serves as an AI consultant across diverse industries, leading workshops and guiding companies from the conceptual stages of AI implementation through to final deployment.
As a guest lecturer, Nicole shares her expertise in Python, machine learning, and deep learning at various universities. She is a regular speaker at renowned AI and Data Science conferences, where she conducts workshops and educational sessions. In addition, she is an influential voice in the data science community, regularly reviewing books in her field and offering her insights and critiques. Nicole is also the author of the well-received online course, “Math for Machine Learning.
Download Event PDF
