Joseph M. Email & Phone Number

San Francisco, California, United States

Developed, designed, and implemented a cutting-edge pure Python model interference framework, empowering 50+ data scientists with a unified codebase for seamless production serving. The framework won an Engineering Innovation Award from Indeed CEO (Chris Hyams).Designed and implemented a secure CIS20 compliant infrastructure, enabling data scientists to analyze highly sensitive demographic user data, including information related to race, gender identity, religion, and other protected attributes.Supported multiple AI Ethics projects at Indeed, enhancing the organization’s understanding of user behavior while upholding the highest standards of privacy, data ethics, and demographic data compliance.Spearheaded the seamless transition of services from a private Austin data center to AWS within a year, considering data volume and resource requirements to accommodate the organization’s growing data needs.

San Francisco, California, United States

Lead revenue optimization teams in San Francisco and Helsinki, Finland, driving collaboration and maximizing revenue performance by rewriting the previous architecture across geographically dispersed teams.Managed backend ads service migration EC2 to GCP (Google Cloud Platform), including the re-coding of dependent services from Amazon services such as S3 to GCP-specific versions like Cloud Storage with ZERO operational disruptions.Enhanced ad auction performance by collaborating with multiple teams to optimize eCPM (cost per mile) and fill rates, resulting in an increase of 8%+ additional ad revenue and improving user experience.Restructured Unity's data ingestion for the analytics pipeline, increasing data handling capacity to 80k-110k requests per second and improving data processing efficiency 35%.Implemented a scalable ingestion system for analytics events across 3 EC2 data centers using Kafka, Spark, and custom Golang services, elevating processing speed by more than double and enabling seamless high-volume data management.

San Bruno, California, United States

Designed and implemented an anomaly detection tool to detect abusive/spammy videos which were on YouTube using causal impact. https://goo.gl/7VjJmeCreated several automated monitoring tools for blacklisting, captcha & de-spamming services.Designed a SOX compliance test framework for Youtube’s Ad monetizer.

San Francisco Bay Area

Designed and implemented the entire crawling & indexing system for Applause. Crawling system pulls all reviews from various App stores (Google, Apple, Amazon, Microsoft, Nook) and analyzes the text (using NLP) to extract what users like and dislike to identify bugs.Applause analyzed 2+ Million apps and over 100 Million reviews. The project was so successful UTest pivoted to making it the main product and renamed the entire company to Applause.

San Francisco Bay Area

Developed the base implementation of WebDriver for Chromium. This involved not only implementing the REST API for WebDriver but also adding all missing functionality to Chrome’s IPC API.Developed a fully automated harness for running clustered machine tests on EC2 along with advanced image comparison algorithms for fuzzy template matching. The system was also integrated with the Ocean book scanning service to provide the OCR for 40+ languages.Developed a pipeline to crawl data on a user's post and personal search history and create a statistical search model. The crawling had to be scalable and follow numerous security and privacy policies with auditing.Used Petacat and Twitter feeds to match queries by interest and reorder search results through the use of an extension with Chrome.

San Francisco Bay Area

Tech lead for the design and development of the “new indexing pipeline” tests. Mentored and trained several new engineers and prioritized all testing requirements for the team. Approved the designs of all components.Lead test developer for the Pinax project. Created multiple performance tests for throughput, number of transactions per second, and latency concerning Pinax observer events.Created a unit test framework, Mr.Unit, for MapReduce to be used by the developers of search for verification of the index pipeline.Developed a server to sample RPC completion times and data transfer size to monitor the indexing pipeline for problems. The tool used a Bayesian model to highlight anomalies while the pipeline was running.Developed tests to benchmark the performance of the Google Account SystemDesigned and wrote all integration tests for the migration of Certified by Google and Orkut to GAIA.