Building a highly scalable ad serving infrastructure
Designed and implement ML data pipeline using ECS, S3 and Dynamo DB to optimize performance for ad programs such as prime video and audio ads.
Design and develop multibid feature for the Ad pacing engine which increased impressions volume by 5%.
Design and develop a high volume caching service using Redis, Kinesis, Kafka and ECS to vend complex time series data to simplify monolithic pacing service.
Developed and launched a new feature called ”Simplified Pacing Profile” which streamlined Amazon’s pacing profile options, resulting in a 1% increase in revenue.
Develop experimentation feature that allowed ML engineers to design and test models for understanding and targeting anonymous traffic, in response to iOS privacy changes, resulting in improved conversion rate of ad campaigns.
Mentor and develop new engineers on the team.
New York, United States
Designed and Developed a service using Elasticsearch and AWS Services such as SQS, S3 etc to ingest 1M emails a month to improve manual invoice reconciliation time from a week to 1 day.
Developed an improved automated invoice reconciliation system that had a 30% improvement over the legacy system using AWS Neptune, SQS, S3 and Elasticsearch.
Productionized a Sum of Subsets algorithm, by using Dynamic programming and a K-SUM algorithm that improved invoice matching by 25%.
Migrate from Kakfa to Redis to reduce on-call load by 20 percent.
Migrated Firetablet Ad Serving from old legacy system to a newer more reactive system that is projected to increase revenue by 66M dollars.
Greater New York City Area
Wrote scripts to convert multiple annotation formats to MSCOCO that scaled to hundreds and thousands of images using Python and Docker.
Built WebApp in Flask, Postgres, React and Docker to query metadata of up to 100,000 images which could scale to millions of images
Worked on Android App that ran on security camera to run various Computer Vision models such as falling detection and object detection.
2019 — 2019
Greater New York City Area
Interned at AI Reverie and worked on a social good project.
2018 — 2018
Seattle, Washington, United States
Built distributed Job in Java to move 10 million records on a weekly basis to Elasticsearch using Bulk API from Oracle Database.
Created service using Java to power a dashboard built in React/Typescript for financial analysts to streamline reconciliation of receipts to invoices.
Built search engine in Java and Elasticsearch to quickly and efficiently search customer information.
Education
2015 — 2019
Rutgers University
Bachelor of Science (BS)
2015 — 2019