Misata is an innovative synthetic data engine that leverages large language models and vectorized NumPy technology, enabling users to generate realistic multi-table datasets through natural language descriptions. This tool eliminates the need for manual schema writing or training data, automatically generating relationally complete data with business constraint support. It supports various large language models including Groq, OpenAI, and Ollama, capable of generating hundreds of thousands of rows per second and handling large-scale datasets with 10M+ rows. Misata provides both command-line interface and Python API support, allowing users to customize data rules and add noise to make generated data more closely resemble real-world scenarios. For developers and data scientists needing test data, machine learning training data, or simulated data, Misata offers a powerful and efficient solution.
