AI in Streaming DAQ*

Streaming Readout has been adopted as the paradigm of data acquisition (DAQ) at many major nuclear physics experiments at LHC, RHIC, and the future EIC. Distinct from the traditionally triggered readout, streaming DAQs rely on modern digital data processing for large factors of data reduction, which opens unique opportunities for the application of AI/ML that is high throughput, low latency, energy-efficient, and reliable. In this talk, we will discuss an array of AI/ML applications for Streaming DAQs on the platforms of ASICs, FPGAs, and novel AI accelerators.