Prateek Shrivastava
About Prateek Shrivastava
Prateek Shrivastava is a Data Analyst at Coupa Software with extensive experience in data science and engineering roles across various companies in India and Ireland.
Current Role at Coupa Software
Prateek Shrivastava is a Data Analyst employed at Coupa Software. He has been contributing to the company since 2018, operating out of County Dublin, Ireland. His role involves leveraging various analytical techniques and models to support diverse data-driven initiatives within the company.
Professional Experience in Data Analysis and Automation
Prior to his current role, Prateek worked at Amdocs as an Automation Engineer, from 2016 to 2017, based in Pune Area, India. Before Amdocs, he held the position of System Engineer at Tata Consultancy Services between 2012 and 2016, also in Pune Area, India. He was responsible for system engineering tasks and various project implementations.
Academic Background in Data Science and Information Technology
Prateek completed his MSc in Computer Science with a specialization in Data Science at University College Dublin from 2017 to 2018. Prior to this, he earned his Engineer’s Degree in Information Technology from LNCT, RGVP Bhopal between 2008 and 2012.
Data Science Projects and Models
During his career, Prateek has developed a range of data science models and projects. He created a Super learner classifier from scratch and applied it to Zalando's MNIST fashion dataset. He also developed a Neural Network model specifically for predicting credit card fraud detection, performed text analysis on New York Times articles using a K-means clustering model, and leveraged ESD and isolation forest models for anomaly detection in load balancer logs.
Implementation of Statistical Methods and APIs
Prateek has implemented the Holt-Winters statistical method on customer time series data for predicting deployment uptime, visualized in Grafana. He integrated Pingdom and New Relic APIs with an Influx database to store and visualize real-time alert data. Additionally, he applied classification algorithms to categorize customers by usage and login frequency and determined the likelihood of abnormal events during instance upgrades or migrations using historical data.