Worked to automate mobile app user experience and accessibility testing. Spearheaded several projects:

Mesmer Gherkin: a compiler for English-language descriptions of app tests (Rust)

• Built a compiler for a test description language similar to English, empowering customers to precisely define app tests, across Android and iOS, without technical knowledge

• Unified many of Mesmer's different internal representations of test logic into one common syntax tree

• Collaborated with the Customer Success Team, writing Gherkin code to rapidly solve client problems

• Employed coverage-guided fuzzing and extensive snapshot testing to find new bugs and stop regressions

The Mesmer Sidecar: instant accessibility audits for Android apps (Rust, Android/Java)

• Built system to locally audit Android apps for accessibility issues, attaching to target apps at runtime and evaluating a proprietary ruleset against data extracted from Android APIs and in-house heuristics

• Built system to attach to a running Android app, extract accessibility data, and run in-house heuristics

• Created WCAG 2.0 rules engine to find accessibility issues and explain what's needed to fix them

• Designed marketing materials, website, and pitch video for release on Product Hunt (#3 App of the Day)

The Mesmer CLI: stable command-line interface to Mesmer tools (Rust)

• Created a tool to expose most Mesmer platform functionality to developers working from the command line.

• Handled frequent changes in unstable internal APIs to maintain a stable ``evergreen'' public scripting interface

• Implemented CI/CD and a comprehensive test suite, auto-importing user error reports into failing tests, and deploying fixes across Mac, Windows, and Linux auto-updaters in 20min

• Provided front-line support, dealing directly with five Fortune 500 clients and many other smaller businesses

Researched both traditional and machine learning approaches to algorithmically determine the physical scale of satellite images. Final model achieved an r-squared of 0.73 using feature extraction and a small dense neural net.

Created a client-facing customization tool for web widgets. This project was deployed to more than 60 local newspapers across the US, integrating with real estate platform enclosure.io and agricultural classifieds site agrisearch.com

Responsibilities included:

• Designing and building full-stack web applications (React/Redux frontend, Express.js backend)

• Creating an automated CI/CD pipeline to build, test and deploy the application to Kubernetes.

Business Impact:

• Allow non-technical users to easily integrate their Enclosure/Agrisearch listings with their own web properties.