Create Ensemble method for preexisting motion prediction model

This project is a continuation of the kaggle completion whose description can be found here: (https://www.kaggle.com/competitions/lyft-motion-prediction-autonomous-vehicles/discussion). There will be an attached GitHub repo that already uses pertained models for the competition task of predicting 3 possible trajectories for self driving cars. For this project I am looking for someone to work with the partial training set as provided and create a ensemble model combining the models Xception41 and Xception65 with ResNet using k means methods returning 3 hypothesized trajectory predictions for the car agents.

GithubRepo: https://github.com/destin168/kaggle-lyft-motion-pred