Time-series clustering of jobs performance behaviors using spark

I am working on a huge number of timeseries of jobs performance behaviors the goal of my work is to cluster a large number of jobs performance behaviors saved as timeseries and identifying the optimal number of clustering using kmeans and Hierarchical and Dynamic Time Warping and compare between the three techniques using pyspark I need someone who has good experience in pyspark especially the dataframe API to help me: 1-convert unsupervised machine learning codes (python code) to pyspark code 2-Submitting an application to YARN on cluster mood using you need to be familiar with Hadoop ecosystem and spark