Open Source AI Data Infrastructure

Pixeltable unifies data storage, versioning, and processing in a declarative interface, eliminating months of infrastructure work. Unlike traditional solutions, Pixeltable handles multimodal data and provides automatic incremental updates.

Pixeltable supports a wide range of applications including RAG systems, computer vision pipelines, multimodal AI applications, and production LLM services. Check out our example applications on Hugging Face.

Computed columns are automatically generated based on expressions or functions applied to other columns. When you add new data or modify existing data, Pixeltable automatically updates these columns.

Yes! You can create User-Defined Functions (UDFs) using the @pxt.udf decorator.

Pixeltable supports a wide range of data types including. See all kinds of types and formats here.

Pixeltable uses lazy evaluation and incremental processing to handle large datasets efficiently. It only loads and processes data when needed and maintains indexes for efficient querying.