TileLang was proposed in April 2025 by a research team from Peking University, Imperial College London, Microsoft Research, and other institutions. The related research findings were published in a paper titled "TileLang: A Composable Tiled Programming Model for AI Systems".

TileLang is a general-purpose block programming model for writing AI kernels more efficiently. This model decouples the scheduling space (thread binding, layout, tensorization, and pipeline) from the data flow and encapsulates them into a set of custom annotations and primitives, allowing users to focus on the kernel's data flow itself while leaving most other optimizations to the compiler.