AI for Quantum Control and Measurement: In-Situ Learning and Adaptive Optimization in QubiCML

As quantum processors scale, maintaining stable and high-fidelity operation becomes increasingly challenging due to drift and variability in qubit behavior. In this talk, I will discuss how artificial intelligence can help quantum systems adapt and self-optimize in real time. Building on the QubiCML framework, I will first describe our implementation of FPGA-based online learning that allows neural networks to update their parameters directly on hardware during qubit readout, compensating for device drift without host intervention. I will then present our exploration of reinforcement-learning-based gate fine-tuning and time-series-driven discrimination approaches that operate in software alongside the QubiC control stack. These AI-driven methods collectively point toward a new paradigm of autonomous calibration and adaptive quantum information processing, where control and measurement dynamically co-evolve with the hardware.