Synthetic Data

Generate synthetic datasets from schemas, real data, or LLM-driven personas. Tabular fit-and-sample (SDV: GaussianCopula, CTGAN, TVAE), Faker/Mimesis schema generation, deterministic PII swap, LLM-driven real-to-synth conversion for unstructured records, and SDMetrics-based quality/privacy evaluation (plus embedding-based leakage checks for text).