Developed robust systems for content moderation, fraud detection, and abuse prevention.

☛ Implemented both near-real-time and batch automated detection systems to protect customers from scam which identified and addressed over 1 million fraudulent entities.

☛ Developed a single-source-of-truth entity state manager using Postgres and DynamoDB which enabled flexible treatment strategies, and improved scalability and auditability.

☛ Led a small team of engineers in collaboration with product manager to build an internal tooling platform using FastAPI and Retool to standardize agent workflows, empower agent investigations, and enable overriding automated actions.

Developed and maintained Spark and Hadoop data pipelines responsible for preparing advertiser data for serving.

☛ Successfully designed, developed, and promoted a unified image crawling and processing solution for all of Microsoft Ads. This service is leveraged by new ads initiatives and enabled experimentation for data science teams.

☛ Engaged in collaborative efforts with partner data science teams on multiple image enhancement projects, spanning from experimentation to production phases.

Developed and maintained Spark and Hadoop data pipelines responsible for preparing advertiser data for serving.

☛ Led initiatives to scale the Image Crawling & Processing infrastructure, enhancing robustness and reliability to efficiently process 100TB+ of data and 100M+ events daily. This effort also reduced the E2E latency by 76% and reduced storage costs by 40%.

☛ Initiated and led the development of a React & ASP.NET web application for tracking Product Ads status and metadata which drastically reduced support investigation times by 50%.

As a course grader, I helped grade assignments and exams of undergraduate students in the Intermediate Data Structures & Algorithms course.