Maciej Ceglowski <maciej@ceglowski.com>
Authorized to work in the United States and European Union. Interested in offsite contract work in any of the following areas:
Built an online book reader for Otworz Ksiazke, a Polish Creative Commons project at the University of Warsaw.
Designed and built an internal data warehouse for Twitter.
Redesigned backend architecture for Aupeo.
Designed and built a spreading activation geographic recommendations engine for Platial.
Overhauled and expanded ad server infrastructure at SixApart.
Built a statistical language guesser for Sphere.
First engineer hired at Yahoo's internal startup incubator. Worked as a backend developer, community manager and bug squasher on the Pipes visual Web programming environment. Designed and implemented a visual airfare search tool called FareMaps.
Designed and built a JavaScript web application for bookmarking New York City restaurants. Created an iTunes recommendations engine using principal component analysis for France Telecom. Co-founded and wrote codebase for Podlinez, a service for listening to podcasts over a regular phone line (sold in December 2006). Built recommendations engine for upcoming.org
Worked in the Research in Information Technology program. Helped solicit and evaluate grant applications, oversaw and attempted to rationalize large academic open source projects (Chandler, Open Courseware, Fedora).
Built a number of open-source search engines and visualization tools for university use. These included a latent semantic search engine, a set of Perl modules for spreading activation search, a part-of-speech tagger for English, unsupervised clustering tools, and a weblog census.
Built database-driven websites for small retail businesses in Vermont.
Most of my work has involved LAMP and AJAX web applications, with some smaller projects in Java and C++. I maintain the following Perl modules on CPAN:
I also run the following online projects:
Pinboard, a lightweight bookmarking site targeted at people with large bookmark collections.
Wrong Tomorrow, a site for tracking predictions by pundits and public figures.
(with Joshua Schachter, Peter Gadjokov)
LOAF is a way to share social network information through email without exposing private information. LOAF files are Bloom filters (a kind of one-way hash) that ride along as attachments to email and can help recipients identify new correspondents who have exchanged email with people they already know.
A statistical language identifier derived from TextCat. Languid uses a statistical algorithm to select candidate languages from known training sets, and narrows its guesses based on which Unicode blocks are used in the unknown text. Technorati uses a version of the Languid engine.
(with Jason Williams)
A set of Flash extensions to Google maps for displaying very large data sets.
A free online directory of bedbug reports at hotels and street addresses across North America.
B.A., Middlebury College, 1997, summa cum laude (double major in Russian and Studio Art).
English, Polish, French, Russian, Spanish