Contributed to the engineering behind synthetic data generation in Microsoft Foundry. The feature lets users generate artificial datasets that mimic the statistical properties of real-world data — useful when production data is scarce, expensive, or sensitive.

The main use cases are data augmentation (expanding training datasets for more robust models) and testing/validation (stress-testing models under varied scenarios without needing real-world data). Under the hood, it uses LLMs to produce predictions across datasets while preserving statistical properties.

Built at Microsoft as part of the Microsoft Foundry platform.