Sean Mccullough Email & Phone Number

San Francisco, Ca, Us

San Francisco, Ca, Us

Mountain View, Ca, Us

Senior software engineer (TL, sometimes TLM) on Chrome Developer Operations. Most recent project in Chrome Dev Ops: CABE (https://github.com/google/skia-buildbot/tree/main/cabe). Designed and led development of a service (gRPC, Go, GKE) to analyze data collected from controlled, randomized benchmark experiments which compare perf metrics from builds at different git commits. The large variety of distinct hardware/OS configurations in the device pool, combined with the large variety of different types of benchmark test cases used across Chrome projects, all contribute extra dimensions to the measurements and complicate nearly every aspect of the analysis problem that CABE solves. Impact: CABE has dramatically reduced the noise (false positives) in reported perf regressions, saving developer hours and machine resources that were previously spent chasing unreproducible regression results.Previously in Chrome Dev Ops: instrumental IC for launching bugs.chromium.org- Python, AppEngine, MySQL, TensorFlow- Designed and built the comment spam filter (Python, TensorFlow): training data pipeline, feature extraction, binary text classifier model, model inference serving- Online schema changes for replicated MySQL db. Enabled prod updates without downtime.- Worked with visiting PhD student to build a bug label recommender (training data pipeline, text classifier model, production serving, frontend logging for accept-clicks)Previously at Google: G+ user analytics- data pipeline for controlled, multi-arm experiment analysis (Go, map/reduce, protobuf): designed and built user map/reduce pipeline to generate aggregate dimension and metric tables from user session event logs.- dashboard for experiment analysis (Java, JS, HTML): Internal UI to render confidence intervals for experiment metrics.- logging instrumentation library for visual elements (JS, HTML): declarative logging for hierarchical UI elements, to improve coverage of logged user events across G+ web UIs.

Mountain View, Ca, Us

The Google.org Fellowship program enables Google employees to apply to work full-time pro bono for up to six months with non-profit grantee organizations addressing problems in areas like education, criminal justice, or economic opportunity.The grantee I worked with during my Goolge.org fellowship was Tech-Ed nonprofit and Google AI Impact AI Challenge winner, TalkingPoints (https://talkingpts.org/blog/googlepressrelease/1/). My role for this project was software engineer. I designed and helped build out infrastructure to support AI applications to improve student outcomes.My technical contributions included: Extracting and transforming unlabeled training data from production systems and building a process to label it via human eval (Node.js, SQL, Google Cloud Storage, Google Sheets, Data Labeling API), creating a data warehouse for analyzing and controlling access to event logs and training data sets (BigQuery), building a text classifier model to assign user categories for targeted in-app feature promotions (Google Cloud AutoML Text), serving inference requests for the classifier model at runtime (Node.js, AutoML API), instrumentation and analysis for online experiments to optimize ranking for in-app promotions across web and mobile clients (Heroku, Node.js, Google Analytics, FireBase A/B Testing).

Austin, Texas, Us

Developed and deployed private-label blogging and online community software for high profile news organizations. Prototyped, designed and built embeddable javascript widgets for online community features including comments, ratings, abuse reporting, product reviews -all with a focus on lowering customer implementation costs and ensuring service scalability. Developed community management workbench features to support editorial control over user contributed content. My work involves everything from brainstorming to use case models to code. C#, Monorail, db4o, AJAX, JSON, cross-browser JavaScript, Atom/REST, IIS, SQL Server, NUnit, NVelocity, NEverythingElseThatAlreadyExistedInJava.Net

Indianapolis, Indiana, Us

Designed and built a procurement workflow system for Telecom service providers and their clients. Modeled and implemented business processes for service and equipment transactions involving multiple partners. Prototyped a datamining application for building social network models from call detail records. Fuego workflow engine, JBoss, Oracle PL/SQL, JSP, J2EE

Reduced implementation/technology risk for AVLabs portfolio companies by providing technical expertise during formative stages of product development. Client companies included * ForwardVue - supply chain risk management system that modelled Value-at-Risk for embedded options contracts * SANNewCo - SAN management software * VeloSolve - IT project management software * Solomio - IVR for wireless customers * eCityLimits - Procurement and asset tracking for data centers and their customers

Established and maintained the technological vision for the company. Raised $1M in funding from Austin Ventures/AVLabs. Wrote code, hired developers and delivered on a $500k technology license.

Worked with early versions of Java to produce a peer-to-peer instant messenger. Also wrote pre-J2EE web app for managing Bot services exposed to the Activerse Ding! IM protocol.