I've worked on a range of projects as a freelancer. These include:
* Collaborating with Yale University scientists to create a visual interface for exploring brain-related genes based on quantitative data about their expression by type of cell.
* Working with Google Fiber engineers to analyze data and design/build an exploratory tool to help them identify what makes wifi slow
* Advising a Kickstarter finance analyst on revenue forecasting analysis
* And, Working with the team at Stamen on 3 projects
The Stamen team has created a fantastic project called Amazonia Under Threat, as described here. After the project ended, there was a need for an additional interactive page for comparing the various types of human activities that affect the region. I built this addition, and wrote about the design process here.
I worked with the Stamen team on Atlas of Emotions, including defining the shapes of the emotions and developing the "calm" page. You can read more about my, Eric, and Nicolette's work defining the emotional state shapes in my essay on The Shapes of Emotions. Additionally, Eric Socolofsky described our team's collaborative and creative process for Finding Calm in the Atlas of Emotions.
This third project was exploratory, rather than explanatory.
Stamen collaborated with scientists and technologists at the Banfield lab in Berkeley to revamp a tool for exploring and analyzing data related to the genomes found in sampled ecosystems (across many tiny organisms).
To me, one of the most important aspects of the project was subtle design decisions that reflected the nature of the data and the types of data questions. For example, in many cases, the difference between 0 and 1 was VERY important while the difference between 10 and 30 was much less important. Unlike many bar charts, the tallest bar is often not the most important. So, in the "bar chart" view shown at left, I encoded values over 10 as the same length as 10 with a pointy top to indicate that it continues above 10. That way super high value bars don't distract as much from the overall chart and smaller values. Also, the light colors for 1 are very bright against the dark background in this color scheme. This way 1's stand out, and gaps (0's) are visible as you scan either bar chart or matrix view.
CEO of Stamen, and design technologist, Eric Rodenbeck described the project in more detail here.