Cloud SIEM

Technical lead for Unity Monetization Trust & Safety department. Led eight engineers across two teams on the technical direction of ads moderation and publisher facing UI tools.

Worked with engineers, product managers, and UX designers, delivered Ad Controls, a tool to increase transparency of ads for publishers, and allow them to granularly filter and block ads from being served. Utilized React and JavaScript for the frontend. Golang, Kafka, MongoDB for the backend. SQL, python, Druid for the data query layer. Node.js, playwright for rendering ads.

Developing a data lake of ads for feeding into automated ML to detect explicit content in ads, utilizing open source models, in order to improve KPI of percentage of ads that have been reviewed, via automated or manual detection.

Built a fault tolerant ad auctioning system, utilizing Golang, Redis, Aerospike, Kafka, MongoDB, handing over seven billion ad requests a day. The auction system is a core system at Unity, and processes hundreds of millions of dollars in net revenue a year.

Developed and maintained a general purpose boolean logic rule engine for filtering incoming and outgoing ad requests, handling over five million queries per second across the service. Written in Golang, utilizing compressed bitmaps, BTrees, Radix trees.

Designed and developed an AB testing system for segmenting traffic and labeling events within the ads-auctioning backends, for evaluating revenue ¬and user retention impact.

Optimized expensive parts of Golang ad serving infrastructure using profiling and code analysis, saving millions of dollars in recurring infrastructure cost a year.

Backend Engineer. Building systems using Play framework, Ruby on Rails, Scala, Solr, ElasticSearch, MongoDB, Druid, MySQL, AWS, and Puppet.

* Prototyped and completed to production a Scala based backend that integrated viewability data from MOAT into our ad campaign creation UI within the first few weeks of working at BrightRoll.

* Ported an existing legacy single node Solr installation to a multi-node fault tolerant ElasticSearch backend.

* Created an API layer that served as a broker between BrightRoll's UI platform and several internal APIs.

* Built a horizontally scalable (in terms of disk space and query time) backend OLAP query service backed by Druid to replace a non scalable single node existing service that was built in MySQL, enabling faster queries and storage only limited by the cost of additional servers.

* Gained operational experience maintaining a Flume / Hadoop / HDFS / HBase data pipeline that ingested over 10 billion events per day, processing 12TB+ of data (uncompressed) across three datacenters.

Designed and built RESTful APIs to enable Multi-Cloud server management, using Ruby on Rails.