Luís Filipe Brandão

I am a Staff Software Engineer at BuzzFeed where I lead a team building highly scalable public APIs and services.

Before, I led backend engineering on the mobile group at BuzzFeed, was a lead engineer at GoPop and Zeega. Before Zeega and GoPop, I was a graduate student in , worked in text mining in Amsterdam at Textkernel and was a Google Summer of Code fellow.

My current interests are large scale distributed systems, the open web, data mining ethics, internet privacy and open source software.

I'm currently in Brooklyn, NY.

Selected hacks & projects


Bombpopper

[2014] [hack] [news]

Bombpopper lets readers observe how they feel about the news they are reading. Using the Guardian API, we created a live news feed where users are asked to rate how they feel about a news article after reading it. That information is aggregated to report on the overall mood of a reading session.

When used in conjunction with other news sources and ranking algorithms, the data collected with this application can be used to better understand how the widespread use of personalization and algorithmic manipulation in news feeds is changing the news.

Developed at the Hacking Journalism hackathon with James Zhang [dev], Lindsey Wagner [concept & design], and Nick Dynan [dev & design]. Awarded Best UX.

http://www.bombpopper.com/

http://hackingjournalism.com


Data integration system for the Berkman Center at Harvard

[2011] [open source] [rails]

The Berkman Center at Harvard collects a great deal of information about websites around the world, including where they are located, if they are filtered or blocked, if they are connected to oppositional political parties in authoritarian regimes, among other relevant data. Since this information is collected by teams working in different research groups, it is spread among multiple databases with different schemas and formats.

With this project, I created a centralized system that allows to search for information about a website across multiple databases. It helped finding connections between data sets that before were invisible.

I designed and implemented this project in the summer of 2010, through a Google Summer of Code fellowship.

https://github.com/luisbrandao/google-summer-of-code-2010

http://cyber.law.harvard.edu/gsoc/Data_Integration_DB


Untitled sock project

[2010] [arduino] [hack]

Untitled sock project was interactive installation developed during the Mediamatic Dev Camp 2010. Using Arduinos for control and Processing to make a live video projection, we hacked a knitting machine to create a game where three players can coordinate to knit a unique pattern using different colors.

Joint work with Fábio Costa, Pol Pla and Sean Folmer.

http://www.mediamatic.net/168234/en/untitled-sock-project-2010

http://vimeo.com/17227787

http://www.oreilly.de/catalog/haekelnkgger/

Work


Interactive Storytelling Platforms and Digital Archives

Zeega, an online storytelling platform [Zeega is open-source]

GoPop, visual conversation [Get it from the App Store]

Digital Archive of Japan's 2011 Disasters, a multimedia collaborative archive [Open-source on Github]


Interactive Web Documentaries

Austin Music Map, a crowdsourced map of Austin's DIY music venues

Black Gold Boom, how oil changed North Dakota

Curious City, what do you wonder about Chicago that you want WBEZ to investigate?

iSeeChange, crowdsourced climate change reporting

Planet Takeout, the stories behind the Chinese takeouts in Boston

Reinvention Stories, the story of the people of Dayton Ohio

Sonic Trace, storytelling project that begins in the heart of LA and crosses into Latin America


Academic Research & Experiments


Expert finding in question-and-answer web services [2011]

Evaluation of the performance of state of the art Expert Finding algorithms on question-and-answer web services. Project for a graduate course in Information Retrieval.

Mean-shift video tracker [2010]

An implementation of a video tracker that works under illumination changes.
Project for a graduate course in Computer Vision.



Domain Based Classifier Selection [2010]

Exploratory research about using the statistical properties multiple datasets belonging to different domains and show how they can be used to provide a practical guidance for finding the best classification method. Project for a graduate course in Machine Learning.

Workshops & hackathons

2014 Mozilla Festival: Building privacy-aware web interfaces :: London, UK [facilitator]

2014 Hacking Journalism :: Boston, MA [hackathon participant]

2012 Mozilla Festival Workshop: Interactive Storytelling with Zeega :: London, UK [facilitator]

2012 Hotdocs Film Festival workshop: Inventing New Forms of Storytelling :: Toronto [panelist]

2010 Mediamatic Dev Camp ’10 :: Amsterdam [hackathon participant]

Contact