Hi, I’m Connor Gramazio. I work on Alexa Machine Learning at Amazon. In my spare time, I contribute to open source visualization and design projects.
I’m broadly interested in the intersection of design and computing, which inspired me to complete a PhD studying how computation could assist visualization design. I actively seek out projects where I can dive in and combine data science, design, and software engineering.
I always aim to innovate meaningful connections between humans and machines. I’m especially interested in work that can augment human expression, be it through creativity, story telling, or other humanistic interaction.
I'm a strong advocate of accessible design and looking at computer science as a field that extends beyond algorithms alone.
Chromaticity is a tool aimed at simplifying accessible design for visualization. It focuses on simple color palette creation through image upload or color selection in perceptual color spaces, and provides a suite of legibility information including information pertaining to color vision deficiciencies. (use) (source)
Colorgorical is a tool to make categorical visualization color palette design easier, given the frequent difficulties that arise when trying to balance aesthetics with legibility constraints. Large-scale evaluations show that palettes automatically made by Colorgorical are as discriminable and are typically more preferable compared to the defaults included in ColorBrewer, Microsoft, and Tableau. (use) (paper) (source)
d3-cam02 is a D3.js module that defines CIECAM02 and CIECAM02-UCS color spaces. CIECAM02 and CIECAM02-UCS are two of the most perceptually accurate color spaces, which is particularly important for visualization design given that perceptual color differences often encode meaning. (use) (source)
d3-jnd is a D3.js module that allows designers to quickly check whether two colors can be easily differentiated, and considers how color discriminability can shift with changes in size. (use) (source)
Exploring visualization design spaces
I created a collection of hierarchal visualization annotations and developed a prototype to explore the design space for phylogenetic design classification results. (paper)
Classifying interaction behavior
I designed a new feature set based on recent eyetracking advances to classify visualization interaction behavior by mining anonymized cancer visual analysis cursor interaction logs. (preprint)
Usability requirements for cancer genomics visualization
I performed a contextual inquiry to synthesize four common cancer genomics task requirements and then evaluated whether MAGI's interface (see below) supported these needs. My analysis focused on including the full diaspora of cancer research subdisciplines to define and evaluate usability.
I performed a series of quantitative evaluations to identify design practices based on how the layout, number, and physical size of data affects visualization search performance. (paper)
GD3 is a declarative cancer genomics visualization library built on top of D3. (source)