Online Analyses and Real-time Discovery Alerts from LIGO-Virgo-KAGRA in O4 and Beyond

The LIGO-Virgo-KAGRA fourth observing run (O4) is underway and
has reported more than 200 discoveries. The scale of gravitational-wave
searches, inference, and alerts has increased significantly over the last
observing runs. In O4, discoveries for compact binary mergers are being
made every 2-3 days, with alerts being distributed in ~30 seconds of
merger time. Low-significance candidates from both compact binary and
unmodeled searches, averaging ~10 every day, are being distributed for
joint analyses. In this context, I will present an overview of the LVK
alert infrastructure. I will also talk about recent algorithm development
for noise subtraction, search, and parameter estimation using neural
networks in the context of ML4GW, a platform designed to accelerate
training and inference algorithms in the context of gravitational-wave
physics. Here, I will mention recent results using these algorithms and
some challenges faced as we look ahead toward production deployment.