Calm where the stakes are highest.

Since 2025 Jaroslav has been Vice President of Customer Engagements. Four teams, about 40 people, report to him: everyone who contracts, implements, rolls out and maintains the platform for existing customers. His goal is to make that whole side of the company run like clockwork.

He still builds hands-on, too: the internal LLM agents and retrieval tools his teams use every day, designed to run inside a bank's perimeter, where data cannot leave the network.

In 2018 he joined ThreatMark, which protects banks and their clients from fraud with AI, machine learning and large language models. He chose it for its mission: after years in consumer lending, he wanted his work to do more good and to protect people.

Fraud is not an abstract risk. Behind every stolen payment is a person who loses savings, sleep and trust in their bank. At ThreatMark, a release that works means an attack that fails and someone who never finds out how close it came. That is the kind of result he wants his work measured by.

As Head of Delivery and Customer Support he owned the delivery chain at Tier-1 banks, where a security review gates every release, while the company grew from about 20 to 120 people and expanded into the US.

From 2013 to 2018 he worked within PPF Group at Home Credit International, its consumer-finance arm, first running delivery and operations for a core banking system serving several international markets.

As group project manager he then led SAP implementations in Indonesia, India, Kazakhstan and China. The four projects ran in parallel, each with its own project manager, and he was accountable for all of them: their risks, budgets and deadlines. He spent over a year on site in Asia, more than half of it in Indonesia as deputy to the local project manager. Every one of the four implementations was delivered successfully.

He spots what can go wrong before it does, keeps teams from different countries and cultures pulling in one direction, and makes decisions under pressure without losing sight of the whole.

He was building AI more than a decade before ChatGPT made it a headline. His first working AI system, in 2010, was his bachelor's thesis in applied informatics at Masaryk University: a poker player that made decisions over a statistical probability model, optimising risk under uncertainty.

As the sole developer at CineServis he then built and sold two products, including automated label printing certified by Czech Post, and made automating manual work his focus. Since then he has kept learning and building machine learning and AI systems, through every wave of hype and every quiet year in between.