MicroLIA detects gravitational microlensing events in wide-field time-series surveys using a machine-learning classification engine. It can simulate point-source point-lens (PSPL) light curves with adaptive cadence or ingest user light-curve collections to build training sets, compute an extensive set of time-series features (including derivative-space metrics and error-weighted statistics), and perform feature selection. The package includes data-imputation methods, Bayesian optimization of classifier hyperparameters, and utilities to save, load, and visualize models and performance. MicroLIA outputs candidate events intended for further vetting by modeling tools such as pyLIMA (ascl:1906.022).