NeMo Data Designer

NeMo Data Designer is an open source framework by NVIDIA for generating synthetic datasets, either from scratch or from existing seed data. It goes beyond simple LLM prompting to produce data with diverse statistical distributions, meaningful correlations between fields and validated outputs.

Data Designer generates text, structured data and images using samplers, large language models, image models or seed datasets, and can build multimodal workflows with image, audio and video context. Relationships between fields are controlled with dependency-aware generation, and outputs can be validated and scored with Python, SQL, custom validators or LLM judges. Generation can connect to external tools through MCP servers, be extended with plugins, and be previewed, resumed and monitored from small experiments to large runs.

Data Designer is installed with pip and works with model providers such as the NVIDIA Build API, OpenAI or OpenRouter. It is distributed under the Apache License 2.0 and actively developed on GitHub; version 0.9.5 was released in October 2026.