Patrick Keenum

Served as a principal software engineer with a front-end focus on developing a blockchain-powered MMO space strategy game, requiring deep technical problem-solving in 3D graphics, WebGPU, and high-performance computing. The game featured 250,000 dynamically rendered asteroids with procedurally generated terrain, each with up to 1.7 million selectable plots, real astrophysics, and true-to-life time scales, necessitating chunk-splitting, dynamic level-of-detail (LOD), and heavy memory management to ensure smooth performance.

I developed the progressive web app (PWA) architecture, utilizing WebGPU, Three.js, and custom shaders to deliver a high-fidelity experience directly in the browser. This involved leveraging web workers, offscreen canvas rendering, and parallelized computations to optimize performance.

I collaborated closely with the art, UI, and back-end teams, integrating design assets, refining UX, and ensuring seamless synchronization with blockchain-based game logic. I also played a key role in debugging, profiling, and optimizing device performance, ensuring a smooth and engaging player experience.

Developed a real-time matching system to salvage otherwise discarded vaccine doses in the initial COVID-19 vaccine rollout by connecting providers with eligible recipients.

The system allowed pharmacies and vaccine providers to list extra doses that would need to be used before expiration due to the short shelf-life of open vials. It then used geolocation-based matching to identify the closest eligible recipients, prioritizing those in the highest-risk groups based on their eligibility phase. It factored in current driving conditions and estimated arrival times to ensure recipients could reach the vaccination site in time.

Developed a data-driven election simulation platform that modeled voting trends using census data, historical precinct-level voting records, and geospatial data. The system analyzed demographic shifts, historical turnout, poll wait times, current COVID-19 data, and early/main-in voting patterns to simulate potential US election outcomes at a granular level.

The project required large-scale data processing. To handle the massive dataset efficiently, the system utilized distributed computing techniques, optimized queries, and in-memory data processing. The front-end visualized results interactively, allowing users to explore different simulation parameters and see their impact on projected outcomes.

Created a marketplace for HR service providers to interact more productively with corporate buyers. Features included a Quora-like Q&A board, comprehensive vendor directory, media dashboard, integrated sales tools for our outbound team, chatbot search tool for corporate buyers, and more. Grew the business to $200,000 ARR within six months of launch.

Constructed a multitenant sales automation platform to manage outbound lead generation for SMB clients. Built tools for prospecting, running multivariate data reports, designing tokenized HTML email campaigns, and managing campaign responses. Utilized machine learning to increase email deliverability and semi-automate prospecting and inbox management over time. The business grew to $1 million ARR.

Helped redesign a fun single-page application used to track and sharing photos of the Despicablimp: a cartoon-shaped blimp that traveled the country to promote the opening of Despicable Me 2.

Developed a video submission contest for Mother's Day 2013, then reskinned it and repeated for Father's Day. Partners sponsoring the projects included Vizio, BestBuy, DreamWorks, and Walmart.

Created an homage to the old-school "Snake" game that came built-in on pre-smart cellphones. This was an afternoon project to my first app with React.