London, England, United Kingdom
Harshit is a Senior Applied Scientist working in Amazon Logistics’ Research Science team based in London, with a total industry experience of 10 years. His primary interests lie in applying ML, Optimization, AI, and scientific techniques to solve challenging real-world problems. Currently, Harshit is working on projects aimed at improving last mile supply chain’s strategic and planning decisions.
● Built an ensemble model to predict cancellation of bookings for Inspection with an accuracy of 72% using logistic regression and SVM to manage resources ● Responsible for data analysis, process automation and generating data driven insights to monitor KPIs & deriving business decisions and for streamlining existing processes
● Responsible for "Loss Minimization" Project and achieved a loss reduction of 10% (month-on-month) on overall monthly revenue of the company by dynamic pricing and optimized city-wise discounting. ● Developed model using logistic regression to do customer scoring based on app usage data and personalizing discounts to optimize marketing using Python ● Responsible for development and implementation of dash boards for Profit & Loss and KPI analysis using QlikSense. ● Responsible for development of a model to classify partner hotels in different categories using unsupervised machine learning in R to make effective and efficient business decisions. ● Single Point Of Contact for reconciliation of the cash outflow and inflow at a central level for all the partner hotels.This reconciliation data was further utilized to generate financial statements of the company.
● Delivered a solution using application of supervised machine learning in R to automatically classify customer feedback with a near real-time accuracy of 90% and developed a visualization of the same in d3.js for a major consumer electronics manufacturer of US ● Generated actionable insights by performing text analytics on customer feedback data using R for a major airlines of US ● Developed a tool to scrape social media website using Python as a part of attrition modelling for Wipro's HR Department. ● Completed a Proof of Concept on “Battery Life Prediction” for a large electronics manufacturer using Regression Analysis ● Automated the tedious and strenuous process of risk/non-risk clause classification based from contract documents using Stanford NLP tools