Solving complex and large scale problems from first principles in the fulfillment of large population of orders placed on Amazon’s retail websites across marketplaces.Domain specialization: Supply chain - Order Fulfillment, Simulations & ExperimentationTechnical Specialization: Machine Learning, Sampling, Data Analysis, SQL, python

Areas: Data Science, Statistics, Machine LearningProgramming Language: Python, SQLTeam: Supply Chain Optimization Technologies (SCOT)- Built a clustering based sampling strategy to create a high-fidelity sample(s) for Amazon's order fulfillment simulation systems- Incorporated demand forecast in the samples, improving accuracy and precision metrics for peak periods such as Prime day and Black Friday sales- Conducted correlation analysis for CO2 emissions and parameters such as distance travelled by a packet, packet size, shipping speed etc. as a POC for a key functionality on the Amazon website.

Areas: Software Engineering, Data ScienceProgramming Language: Python, SQL- Developed a predictive model to identify the merchant’s business category - food & beverages, fuel, consumer goods etc., contributing to targeted marketing campaigns and promotional offers.- Built and conducted experiments to verify all the incoming and outgoing payments reconciliation for all product lines, proactively checking potential bottom line impact by preventing money leakage and future cost in audit disputes.- Improved the daily payment settlement ratio by identifying users having dysfunctional accounts for reasons such as blocked/frozen accounts, limit exceeded etc. using pattern recognition in recent historical transactions, enhancing user trust and experience.

- Built a classification model using random forest algorithm to predict payment disputes on open invoices, decreasing the effort in consolidation of payments at quarter end by 30% and reducing bottom-line impact by 25%. - Developed novel solution to forecast demand from customers’ buying trends using croston algorithm and SAP Predictive Analytics Library (PAL), reducing material stock-outs by 95% and adding 15% to top-line. - Developed time series ML model using arima algorithm to predict production order time for a manufacturing client. The prediction results helped in confirming delivery time to customer in material routing, leading to accurate estimation of order delivery and reduced gap between planned and actual.- Implemented Opentext Vendor Invoice Management (VIM) framework to create and process ~15,000 vendor invoices monthly, for a multinational professional services firm. Developed end-to-end invoice handling process including reading of pdf/.doc(x) documents using OCR, exception handling, invoice approval workflow, and invoice posting and storage in SAP system.

- Joined Deloitte as a campus hire- Automated bank cheque payment method for the Canada business unit of a Fortune 500 client. Designed a reusable bank cheque template using Adobe forms in S/4 HANA to dynamically print cheques in English, French, and Español along with MICR code and digital signature capability. Eliminated effort for writing 4000+ monthly paper cheques and ensured timely and accurate payments to vendors.- Restructured the digital procurement platform of a Japanese automobile giant to handle a 25% rise in transaction volume by eliminating multiple IT systems and integrating all the steps of the procurement cycle from request for quotation to invoicing in one SAP system, reducing the sourcing cycle to 5 days from 23 days in the past.- Launched SAPtimes – a service line monthly newsletter - a first in the US India Deloitte offices, facilitating effective top-down and bottom-up communication between service-line leadership and employees by sharing leadership thoughts and latest firm news with all the 2000 professionals across the practice.