Pratik Gujral

➢ Demand Forecasting: Forecasted district wise daily demand for >80% of volume with 73% accuracy enabling proactive logistics planning, and pre-dispatching orders, reducing fulfillment time by 45% for 17k+ B2C dealers ➢ Sales Bot: Forecasted monthly sales volume for area sales officers every day with >75% accuracy for 90% of volume. Provides feedback to sales officers if they will meet their monthly sales target based on current sales ➢ Logistics Optimization (ongoing): Develop LP solution to map open cement orders to delivery vehicles for 18 plants across India, maximizing capacity usage and minimum number of vehicles subject to business constraints ➢ Pre-Journey Planning Automation (ongoing): Manage development of an optimization solution to generate plans for Sales Officers’ physical & virtual meetings with dealers aimed at reducing churn and acquiring new customers

➢ Usage-Based Insurance: Built driving behavior profiles for insurance by performing data cleaning in PySpark on AWS EMR, EDA, anomaly detection, and feature engineering computing events of harsh acceleration, braking, over-speeding, seat belt usage, driving time, road type, and used past accident and workshop repairs data. Trained hierarchical clustering model to classify vehicles into different risk categories used to calculate insurance premium discounts. ➢ Set Up Data Lake and Layered Data Architecture: Architected a unified platform for data analysis by setting up a multi-layer architecture integrating vehicle IoT data, CRM, Workshops repair datasets on AWS and IBM Cloud. Setup Spark jobs for incremental data load into data lake ➢ Dealer Management System Analysis: Identified customer pain points by analyzing vehicle repairs data finding patterns in automotive components that require frequent repair and maintenance ➢ Advanced Driving Assistance Systems (ADAS): Built proofs-of-concept projects and evaluated vendors for camera-based Lane Departure Warning, steering angle and yaw-rate sensor-based Vehicle Sway Warning system, Driver Drowsiness Detection, Blind Spot Monitoring, Surround-View Camera System, Forward Collision Warning (FCW), and Auto Emergency Braking (AEB) ➢ Miscellaneous Projects: • Built a proprietary Python module that uses Spark clusters for common automotive data-wrangling tasks • 30% cost reduction on AWS bills by implementing AWS best practices. Built a system to inform performance metrics for running AWS EC2, EMR instances using AWS Cloudwatch, Lambda, SNS • Received Employee of the Quarter Award for training 50+ fellow employees in Data Analysis using Python (inc. Numpy, Pandas, Matplotlib) and SQL over two months online during COVID-19 lockdown

Built a time-series forecasting model predicting the volume of ATM transactions for over 12000+ ATMs belonging to leading banking firms with over 85% accuracy using SARIMA and Holts Winters in R.

Results awaited