Passive brain-computer interfaces (P-BCIs) offer the ability to dynamically adapt system interactions by inferring users’ cognitive states, such as levels of mental workload, from neural data. The P-BCI devices have many applications from analyzing and designing educational materials to monitoring the mental health of professionals operating airplanes, nuclear power plants, and airline traffic dispatchers. Electroencephalography (EEG) is the most suitable P-BCI device that can easily find applications in daily life due to its cost efficiency and device setup simplicity as compared to other devices like Functional near- infrared spectroscopy (fNIRS) or Functional Magnetic Resonance Imaging (fMRI) that are either costly or are bulky equipment that cannot be used in real life scenarios. The goal of this work is to classify the cognitive load of the user using his/her EEG output using machine learning (ML) techniques, following the MINDSCOPE (Machine Learning Inferencing of NeuroData for Seamless Cognitive Overload Prediction and Evaluation) methodology. The data acquisition was done by recording the EEG signal of the users performing n-back tasks of two different load intensities (rest and 2-back). This study accurately classified user cognitive load as high or low, allowing for comparison of several ML models and their performance in cognitive load discrimination. Feature subset analysis identified the combina- tion with the highest classification accuracy, shedding light on effective feature characterization for cognitive load assessment. Depending on the context of P-BCI use, the current work gives a starting point for choosing appropriate ML paradigms and feature characterizations to incorporate for a specific cognitive assessment task.