The El Niño-Southern Oscillation is a complex, interannual phenomenon occurring in the tropical Pacific characterized by two dominant modes: El Niño (warm state) and La Niña (cool state). As one of the leading sources of climate variability, skillful prediction of ENSO is crucial for societal preparedness due to its widespread teleconnections (e.g., drought, floods, wildfires, and extreme heat). El Niño and La Niña events are often identified using sea surface temperature anomalies averaged across the Niño 3.4 region (i.e., Oceanic Niño Index). Several other ENSO indices, such as the Southern Oscillation Index and the Multivariate ENSO Index, also consider other meteorological variables and/or regions. Although the dynamics of ENSO span from the ocean subsurface through the Earth’s troposphere and lower stratosphere, many indices used to identify and track ENSO onset, materialization, and lifespan only capture a portion of the phenomena’s complexity through variables at the ocean surface. Additionally, regions bounded by fixed latitudes and longitudes may not fully explain different ENSO ‘flavors’ and diversity. To better characterize ENSO and capture its diversity and inherent nonlinearities, this study leverages a neural network called an autoencoder to derive a new index for ENSO. An autoencoder uses an unsupervised learning approach to learn without the need for labeled data. The autoencoder is trained using three fields: sea surface temperature, ocean heat content, and zonal wind stress (which acts as a proxy for westerly wind burst events). In addition to including more fields than previous indices, the region of interest encompasses much of the tropical Pacific region, freeing our derived index from a fixed geographical domain. The data used in this study is from the Department of Energy (DOE) Energy Exascale Earth System Model Version 2 Large Ensemble (E3SMv2-LE). These simulations have 21 ensemble members, with both historical and future runs. We also use the pre-industrial control simulation to examine ENSO within E3SMv2-LE without anthropogenic effects.