Thiago Teixeira

My research interests are human sensing, data interpretation, and system development, aimed toward enabling the massive-scale systems of the future.

Born in Rio de Janeiro, Brazil, I moved to the US in 1999 to pursue my Bachelors degree in Electrical Engineering at Johns Hopkins University, with second major in Mathematics. From 2003 to 2005 I worked at the Sensory Communication and Microsystems Laboratory at JHU, where I did my Masters work interfacing address-event cameras and acoustic localization devices to wireless nodes, under the guidance of Andreas Andreou. In the period of 2005–2010, I pursued a PhD at Andreas Savvides’s Embedded Networks and Applications Lab (ENALAB) in Yale University. There, I developed a system composed of networked sensors with the intent to aid people in their homes and workplaces.

Soon after, I joined INRIA Paris-Rocquencourt working in Valérie Issarny’s ARLES team, toward a smart middleware for the Internet of Things. I, then, Joined Google as a software engineer at Google, more specifically in Carrie Grime’s Infrastructure Quantitative team, learning a lot about data analysis while building some fun visualization tools for internal use.

I am now at Google[x], Google’s applied research lab that specializes in crazy “moonshot” ideas. I am working on something exciting and new — but also very secret. I will keep you posted when we are ready to tell the world what we are up to!

Prior to joining Google I researched embedded systems, middleware, and sensor technologies that enable massive-scale sensing and understanding of people’s actions and interactions within smart environments. This website focuses largely on that work.

As the number of sensor devices such as mobile phones and specialized sensor nodes continues to grow past the billions of units, existing research problems that are currently solved solely on a case-by-case basis by field experts will suddenly become too common to be handled in such an ad-hoc manner. Instead, fully-automated methods must necessarily be devised. In my research I have worked toward automated information fusion in the upcoming massive-scale networks of sensors. The ultimate goal is to achieve a truly plug-and-play system that can be easily extended with any sensing modality by simply buying a new sensor and pressing the “on” button.

Previously, in my Ph.D. and Masters research, I studied the field of human sensing, and developed a sensor network to detect, track, and identify people in assistive home and office environments. In this work, one of my primary concerns was preserving a certain level of privacy, by enabling the use of cameras that extract features but cannot take pictures. I considered the problems of detecting people, uniquely identifying them and making inferences from their interactions with the surroundings. For this, I have developed a wireless network of custom camera-nodes that is able to locate moving people in real time with minimal calibration or set-up. This infrastructure has already been used in real homes to infer activities by reasoning about people’s locations in relation to known objects in a scene. Much of this research focused on problems that arise in multiple-person environments, such as the separation of an unlabeled data trace containing different people into single-person traces. View research projects.