Research on interpretability and control of neural networks, primarily language models, as well as vision-language models for understanding information-rich graphics such as charts.
Research and engineering at the intersection of machine learning, visualization and HCI.
Early contributor to TensorFlow.js, the leading deep learning library for the web. Joined the project before it became part of the TensorFlow ecosystem and eventually led development of the tf.js 3.0 release. Led development of tfjs-vis, a companion library for visualization of in-browser training dynamics and model behavior, as well as tfjs-react-native.
Co-created a podcast, Tic-Tac-Toe the Hard Way, to introduce non-experts to the nuts and bolts of deep learning.
Later projects focused on visualization for generative model evaluation and interpretability for multi-agent RL systems.
Design and development of custom web-based data visualizations using open source technologies.
Founding member of the data visualization team at Bocoup. Worked on a wide variety of client and internal projects, both as a team lead and individual contributor. Responsibilities included client relationship management, technical project planning, user and stakeholder interviews, visualization design, prototyping, and implementation of data-driven web applications.
Published blog posts on visualization topics and developed classes and workshop materials for in-person delivery, including Introduction to D3, Text Analysis and Visualization, and an Art & Code workshop designed to introduce high school girls to computational creativity.
Research and development of novel information visualization systems for external and internal audiences.
Designed, developed, and deployed Many Bills, a publicly accessible web-based visualization of congressional legislation. This work was published at CHI and SIGIR, and the project was also a 2011 Webby Awards Honoree.
Developed event sequence and timeline visualizations that were incorporated into IBM Cognos products.
Mentored two graduate student interns.
Research on visualization in socio-technical systems to support communication and collaboration.
Built and launched a visual microblogging site that allowed people to document their everyday data and create statistical graphs of their lives using a simple Twitter-like text interface.
Developed a novel programming environment to encourage collaboration among distributed programmers within communities of practice by tracking and visualizing the movement of code with very fine granularity.
Gave numerous presentations to sponsors in both small and large groups.
Development of Java-based NextText and TextEngine libraries for dynamic typography and computer-vision-based integration with real-time human performance.
Designed and implemented installation and performance art pieces. Led a team of three developers during my final six months.
S.M. Media Arts & Sciences
B.Sc. Computer Science, with Hons. Minor in Digital Image/Sound and the Fine Arts.
Current:Python, JavaScript/TypeScript, HTML/CSS, d3.js, PyTorch, HF Transformers, Sketch, Adobe Illustrator
Past: Ruby, C++, Java
Conference Reviewing: NeurIPS 2024 (Top Reviewer) | NeurIPS 2025 | ICLR 2025 | ICML 2026
OpenVis Conf (2013-2018): Served on the OpenVis Conf program committee from its inception. Also administered the diversity scholarship program.