AI Engine kernels are computation functions that work together to build data flow graphs. Multiple subgraphs can be stitched together into larger high-level graphs. This beamforming tutorial uses three beamforming kernels that implement matrix multiplication functions. These kernels are strung together to form the downlink and uplink subgraphs. Three uplink and three downlink subgraphs are instantiated in a high-level data flow graph TEST_BF defined in test_bf.h. The AI Engine application contains an instance of the TEST_BF graph and can be compiled and simulated using the AI Engine tool chain.