Apr 2021 – May 2022 · 1 yr 1 mo

• Developed UNIFIED PRICE COMPARISON pipeline using SQL, PySpark and Redshift to measures the overall price competitiveness of products sold on Amazon by third-party(3P) merchants and amazon retail with the competitors and provide actionable insights. I upgraded the entire pipeline to ingest competitor classification data from a new tech system, resulting in a 5% increase in overall price competitiveness.
• Created S3P-ONCALL-DASHBAORD for real-time monitoring of metrics like disk utilisation, table statistics, job run status, Spectrum and Cradle costs for the entire S3P org. By tracking the complete upstream and downstream lineage, it provides alerting capabilities by spotting failed job executions. The tool saves roughly 1.5 hours each day by eliminating need for manual tracking of more than 500 production jobs for failures and highlighting actionable items for on-call.

Jan 2020 – Mar 2021 · 1 yr 2 mos

• Built BEST SOURCING STRATEGY (BESSY) tool using SQL, Redshift and QuickSight based on decision tree classification model that optimises the procurement of goods by selecting the best inbound sourcing strategy for vendors and manufacturers across EU. BESSY resulted in lowering procurement costs and increasing inventory availability, resulting in faster delivery and lower price points for customers. This project contributed to a 10% increase in overall savings (EUR 2.5 billion) for FY20-21.
• Expanded GLOBAL LISTING service to new marketplaces Netherlands, Poland and Sweden. Global Listing is a vendor self-service solution that allows vendors to expand their current catalogue choices to new marketplaces in the EU region without interacting with a vendor manager or Amazon service. The project assisted in the onboarding of 22 million products from current vendor catalogues into newly launched marketplaces.
• Created FORTUNE tool for publishing monthly Profit & Loss (PnL) reports for multiple businesses across all product categories. The projects aims to automate the entire process of preparing PnL statements by creating an end to end data pipeline jobs.

Jun 2018 – Dec 2019 · 1 yr 6 mos

• Implemented EMR LOG ANALYSIS tool using Python for quickly identifying and visualising issues such as cluster provisioning errors, throttling & system errors, memory/disk/CPU issues and network latency with Elastic MapReduce (EMR) service by analysing Hadoop system logs, Application logs and backend AWS infrastructure logs. The project aided in the reduction of the service team's overall ticket response time from 2.5 days to 1 day.
• Optimizing job flows (Spark/Hive/MapReduce) by analysing YARN and applications logs based on partition analysis, caching techniques, serialization/compression methods, and the YARN resource configuration.
• Designing cloud architectures using open source Hadoop platforms like Hive/Hue, Spark, Presto and Big Data services offered by AWS like EMR, Athena, Redshift, S3, DynamoDB and Glue to meet customer requirements.