We introduce Collie，an online platform to assess the quality of nursing home service powered by machine learning, which enables user to choose a serial of selected algorithms for different demand. Previous research has proposed the point on how to measure the quality of nursing home service. But traditional service relies on manual investigation which present serval problems like the process of rating quality is complex and insufficient. For the above problems, Collie focus on supporting users to assess the quality of nursing home service. Within the service assessment, we develop a serial of algorithms to deal with different dataset. But more importantly, different database has different features, a group of algorithms is not enough to fulfill users demand. Therefore, we propose a system which enables users choose highly competitive algorithms to solve their problem. To support the core idea, we provide a high performance infrastructure features asynchronous I/O, fast and error tolerant data pipeline. Through a series of experiments, we show that our system outperforms the baseline significantly in terms of system performance, algorithm accuracy. Promising result has been retrieved from our algorithm which indices MAE of 0.5 at sample dataset.