Why HASLAB

Inspectability is the point.

Principles

Build from evidence outward.

Physical-AI systems need more than peak arithmetic. They need explicit model conversion, data movement, preprocessing, control, error behavior, and reproducible end-to-end measurement.

Open contracts

Numerical rules and interfaces should be readable, executable, reviewable, and versioned.

Honest boundaries

Unsupported behavior and host partitions are declared rather than hidden behind fallback.

Measured progress

Capabilities change status only when the required evidence is checked in.

People

Designed for serious collaboration.

The project is intended for university researchers and students, open-source hardware contributors, commercial and applied research labs, and teams exploring FPGA or ASIC implementation.