Graham Wright Email & Phone Number

Train and deploy vision-based machine learning classifiers to detect and remediate child abuse and other graphic content across Facebook and Instagram (public announcement: https://goo.gl/3JHGN5)Mentor and ramp up junior engineersWork and coordinate with countless teams across the FB family of apps to expand objectionable content detection to existing and new products and initiatives

Software engineer on Care Machine Learning: Graphic Content within Community Integrity (https://goo.gl/MaUK5b)Training and deploying a wide range of computer vision, text, and contextual based objectionable content machine learning classifiers to keep Facebook safeEngineering efficient systems to manage and serve remediation and scores throughout Facebook and InstagramWorking primarily with Hive, Presto, Python, Hack, and Haskell

• Developed end-to-end pipelines for retraining and deploying objectionable content classifiers.• Improved classifier performance and reduced CPU cost by 25% by enhancing textual analysis.• Scaled text models to use 10x more data in the same amount of timeMain languages consisted of Python, Java, and Haskell/Haxl.

Atlanta, Ga, Us

Equifax is a top credit reporting agency in the US, housing the credit records for hundreds of millions of consumers and businesses._My roles:Successfully completed a proof of concept with Elasticsearch to improve credit reporting search match processes. Capable of a $1m+/year increase in revenue.• Personally installed, configured, administered, and loaded an Elasticsearch cluster with over 150 million records• Performed meticulous data analysis, auditing, and testing in Python to compare hit rate performance between existing platforms and Elasticsearch• Developed and integrated a C++ Elasticsearch API into the back-end of our core credit reporting platform• Performed further optimizations to increase hit rate by 25% and decrease query time by 50%• Authored a very comprehensive wiki with steps to reproduce and productionalizePerformed full stack web development in Google Web Toolkit on an Agile team for business account monitoring.Upgraded Coverity for static code analysis and integrated the analysis process into daily builds for all platforms.Screened and interviewed candidates for full time and intern/co-op software development positions.

Atlanta, Georgia , Us

Comp.Social Lab website: http://comp.social.gatech.edu/Public GitHub: https://github.com/compsocial/CREDBANK-dataBackground: This project tracks the credibility and spread of rumors for events over time on Twitter.Generated and analyzed time series and linguistics features for over 90 million tweets across over 1300 events. Heavy use of Python (matplotlib, pandas, and ipython notebook).ACM Paper: Credibility and the Dynamics of Collective Attentionhttps://dl.acm.org/citation.cfm?doid=3171581.3134715

Atlanta, Ga, Us

Developed encryption utilities in C# for deployment as stored procedures in SQL Server and DLLs on web servers. Integrated the encryption utilities with existing ASP.NET web applications.Configured Zementis Universal PMML Plug-in (UPPI) on Hadoop Hive and GreenPlum for use in new scoring analytics on over 92 million records. Automated the process with bash scripts and improved runtime by over 50%.Deployed a Jenkins instance to support continuous deployment. Wrote bash and batch build scripts to automatically deploy ASP.NET web applications upon Team Foundation Server version control check-in.Developed data audit validation scripts in Perl to enhance existing auditing procedures.Worked on minor defects and enhancements in the backend of an ASP.NET internal report generating tool.Performed resume and phone screenings for full time position candidates.

Scheduled one on one tutoring sessions with students, prepared for sessions by reviewing material from past coursework and learning new material added to the curriculum, prepared examples for students, answered student questions and concerns, and mentored students.

Atlanta, Ga, Us

Worked on an end to end solution for delivery of scores to customers in SQL and Java that speeds up delivery by over 75%.Developed PostgeSQL functions for keying 3rd party financial data to our data, utilizing hierarchical traversals of the database.Interviewed co-op/intern candidates, performed Agile software development, developed a department-wide roadmap of continuous integration improvements, and authored a wiki for future co-ops and interns.

Bachelor Of Science (B.S.), Computer Science

Education record