Suspended optics in gravitational wave (GW) observatories are susceptible to alignment perturbations, particularly slow drifts over time, due to variations in temperature and seismic levels. Such misalignments affect the coupling of the incident laser beam into the optical cavities, degrade both circulating power and optomechanical photon squeezing and thus decrease the astrophysical sensitivity to merging binaries. Traditional alignment techniques involve differential wavefront sensing using multiple quadrant photodiodes but are often restricted in bandwidth and are limited by the sensing noise. In this talk, I will present the first-ever successful implementation of neural network-based sensing and control at a gravitational wave observatory. The talk will cover the recent results of designing multi-input-multi-output (MIMO) controllers using deep reinforcement learning (RL) and the corresponding improvement in sensitivity to compact binaries observed at the GEO600 GW detector.