Lead efforts to design and implement shelf space optimization of the product catalog at real time. This allowed fine-grained control over the contents of the product catalog in the

Experiment and evaluate product categorization using various machine learning models.

Build tools to monitor and evaluate categorization performance

Created visualizations and published metrics for different categorization models using Grafana, StatsD and Graphite.

Supported efforts to create and onboard new revenue stream channels.

Designed and created automated slack alert messages for the revenue stream pipeline integrated with grafana dashboards

Designed a proof of concept based on word vector embeddings to create a related search recommendation engine

Working on design / re-architecture of the machine learning pipelines

Experiment and evaluate catalog attribute classification task using various machine learning models to set the ground up for guided / bubble search based user experience.

Designed , tested and deployed data pipelines for real time feed ingestion that powers everything from the machine learning

Build Flink jobs to do realtime extract, transform and load operation for data consumption by the downstream systems - Created ansible scripts for seamless deployment using Jenkins.

Profiled the JAVA application for memory leaks and optimized the API’s for better throughput

Create and apply algorithmic patches/fixes for both real-time and batch components of the AMEX real-time offer and merchant Recommendation platforms.

Designed a real time shopsmall merchant typeahead, name and keyword search for both secure and non-secure user experience.

Implemented a peer scoring feature for the real time component of the collaborative filtering algorithm to remedy the cold-start problem.

Created ETL’s for the batch component of the US merchant scoring application.

Build control cell logic to assign scores to the merchants based on popularity and random scoring algorithms.

Automated interest, commerce and preference graph computations. Designed a python script to extract data from cornerstone 2.0 repository using a headless browser.

Redesigned the similarity matrix to incorporate algorithmic enhancements

Designed RESTful web applications for shopsmall maps migration and built a location aware search functionality for the maps.

Created scripts to automate the process of continuous deployment in the developer environment with the goal to improve developer experience in the team.

As a part portfolio engineering group within the Digital Offer Ecosystems team (DOE) brought in a few of the best programming Practices within the US scoring team.

Worked on designing data pipelines handling multiple TB’s of data for the personalization and LET re-engineering datasets.

Designed JMeter performance testing scripts for stress testing to identify possible code/JVM parameter optimizations in the JAVA applications. Wrote scripts to scan through GC logs and aggregate statistics to identify performance bottlenecks.