Profile
Background
Tristan Fransen is a Senior Machine Learning Engineer at Pennylane, the financial and accounting software for small and midsize businesses and their accountants. On the Agent Behaviour team, he works on evaluating and improving the behavior of the company’s AI agents. When he joined, he first helped standardize and scale the classical machine learning pipelines, from building ground truth to production deployment.
He has worked in applied AI for more than eight years. In 2018, he joined Shippeo, a real-time logistics tracking platform, as its first Data Scientist, then led the team. There he helped build algorithms that predict shipment delays for large groups such as Carrefour and Renault, covering more than 28 million shipments a year worldwide.
Tristan graduated from École polytechnique.
Perspective
Digital and GenAI credentials
From supply chains to accounting, the core of Tristan’s work stays the same: making sure a model is reliable in real-world conditions. At Pennylane, he builds evaluation pipelines and tooling for agents that handle more than 40,000 user requests a day, in use cases ranging from financial analysis for small business owners to automating tasks for accounting firms. To define what a “good” answer is, he relies directly on the work and methods of accountants. His philosophy: deploying a model has become relatively easy, but the real challenge is proving that it makes the right decisions and gets better with every new version.
