A few key takeaways from his work:

🔹 Data scarcity is not a single problem. Financial institutions need a structured way to decide whether to extend data, share data, use synthetic data, apply federated learning, or combine several of these approaches.

🔹 Synthetic data sharing can become a new form of privacy-preserving collaboration. Instead of exchanging raw transaction data or only model parameters, institutions can share privacy-tested synthetic data that preserves useful patterns while protecting sensitive information.

🔹 Synthetic data in finance needs to move beyond rows and columns. Transactions happen over time and between connected parties, so good synthetic data must capture these patterns too and we need clearer ways to evaluate whether it actually does.

Congratulations again, Fabian: well-deserved! 👏