Seth Falcon

Contributing software engineering expertise to the Program in Computational Biology at Fred Hutchinson Cancer Research Center. Focused on advancing software for the analysis of high-throughput sequencing data for the Bioconductor project and making improvements to R to support the project.

Development lead for Evri's Zeitgeist system that collects, analyzes, and vends usage data. Collaborated with product manager, designer, and engineers to define, estimate, and implement features. Built a fault tolerant scalable system of loosely coupled web services to aggregate and serve time-windowed activity data using HTTP/ReST, Java, BDB JE. The services are used to build popularity-driven user facing features and as a data source to improve ranking results. Initiated PopWhy, a feature to match popular entities with a news story explaining their popularity. Guided the project from prototype to production. PopWhy is used on the homepage, by a Twitter bot, and Evri's iPhone App. One of two engineers that produced EvriVerse, Evri's iPhone App. Self-taught Objective-C. Eliminated memory leaks and significantly improved performance. Dramatically improved Evri's canonical image coverage by creating imageherd, a simple Erlang-based web service and Javascript bookmarklet that allows users to select images from their browser window.

Technical Lead and release manager for Bioconductor, an open source project that provides tools for analyzing high-throughput data arising in the context of biological research. Direct supervision of three staff programmers. Responsible for managing biannual releases involving the coordination of developers distributed world wide to produce over 230 packages. Supported developers and users via project email lists. Added a global string cache to the R implementation that significantly reduced the memory footprint of common bioinformatics data sets. This effort was recognized by an invitation to become a member of the core R development team. Significant contributions to Category and GOstats packages (both written in R) to implement graph based statistical tests for over representation of entities in an ontological database. Improved performance and added features to RSQLite, an R package providing bindings to the SQLite embedded DBMS. http://cran.r-project.org/src/contrib/Descriptions/RSQLite.html Taught advanced R programming courses in USA and UK including OOP in R and C extensions. Designed and implemented weaver, a caching system for dynamic documents used for reproducible research.