Founding team member building Pendo's conversational AI platform. Creating AI-powered interfaces that enable natural language interaction with enterprise data.

• Designed and built extensible artifact rendering system integrating Python backend outputs with Vue components, enabling rapid development of new artifact types across teams

• Implemented full-stack document generation using LangGraph/Langchain, including schema-based templating system for workflow-driven and freeform documents with streaming support

• Built custom @-mention system with Tiptap extensions for entity reference in AI conversations, optimizing performance to handle hundreds of thousands of entities without blocking user input

• Developed markdown renderer supporting custom syntax and LLM response streaming with token-by-token rendering

Founding engineer building note-taking tools for developers. Established state management architecture and built collaborative editing features that resulted in 46k documents created and 7k shared.

• Established state management system using Recoil (100+ atoms) with Redux-pattern action/reducer dispatch for command palette managing 30+ action types

• Built React components for virtualized lists and collaborative editing features, including conflict resolution algorithm for "Suggest Edits" mode preventing simultaneous user editing conflicts

• Led React Native mobile app development using Expo, implementing custom animated draggable virtualized list and managing app store releases

• Shipped high-impact features including single-click document creation workflow

Continued development of Sterling, a web-based visualizer for relational modeling languages. Designed and deployed large-scale user studies to evaluate visualization techniques, then implemented data-driven improvements based on results.

• Designed user studies exploring how cognitive science concepts could improve visualization approaches in Sterling

• Implemented user studies as web applications using TypeScript/React/Redux, deployed via Heroku and MongoDB

• Ran studies with 850+ participants using Prolific and Mechanical Turk, performed statistical analysis in R

• Implemented new visualization modes in Sterling based on study results

• Successfully integrated Sterling with Forge, a tool used to teach formal methods to hundreds of students each semester in Logic for Systems course at Brown University

Developed novel approaches to modeling scientific software using lightweight formal methods. Created Sterling, a web-based visualizer for relational modeling languages, and built visualization tools for large-scale computational geometry datasets.

• Designed and built Sterling using TypeScript, React, and Redux with custom interactive directed graph visualizer and scripting interface for domain-specific visualizations

• Developed models to verify correctness of sparse matrix computations using Alloy modeling language

• Advanced adcirc.io with custom WebGL renderer handling millions of triangles, generic node selection algorithm supporting arbitrary selection shapes, and interactive data exploration tools including node-specific timeseries plotting

• Built web-based visualization and analysis tools for finite element storm surge simulations with real-time 3D rendering

• Wrote data collection and processing software in C++, MATLAB, and Python as part of an engineering team piloting a LIDAR system used to monitor beach erosion

• Developed data collection pipeline, including automated scheduling of LIDAR scans, processing of LIDAR data, and publishing of data to web portal

• Developed GUI for the visualization and processing of wave runup data from large LIDAR datasets using C++ and Qt

• Designed and developed an automated filter for removing noise and differentiating natural environment features such as buildings, plants, and sand from large LIDAR datasets, implemented in MATLAB and C++

• Developed 2D and 3D visualization and animation tools for LIDAR datasets using C++ and OpenGL