Spencer Ferguson-Dryden

Led development on a highly cross functional effort to address a group of high severity “long tail” autonomy scenarios.

Architected a Python gRPC web service enabling vehicle-to-vehicle and dispatch-to-vehicle communication for the entire autonomous vehicle (AV) fleet, meeting near-100% uptime and sub-500ms latency requirements.

Built custom pipelines & infrastructure using Terraform, bash scripts, and Python to ensure high availability, streamline CI/CD, implement tracing & alerting, enforce authorization, and perform database management.

Built an end-to-end high-definition mapping system allowing AVs to remap dynamic areas of the road on the fly. The system enabled AVs to obtain an accurate version of the map to navigate complex environments within minutes of detecting inaccuracies, compared to the prior system that took days while also requiring vehicles to download new map content while at rest in a terminal.

Wrote embedded software modules in C++ to improve autonomous driving performance when encountering complex construction zones.

Reduced fleet-wide operational risk by creating an early warning alert system for remote operators in advance of risky scenarios.

Led development of multiple cloud applications microservices to handle high-concurrency file uploads in support of a multi-million dollar U.S. federal contract

Built custom PHP modules for a high-traffic Drupal-based CMS site critical for U.S. employers to maintain Equal Opportunity certification

Developed a React component library using Storybook.js to streamline the implementation of front end components across multiple federal web applications

Developed an innovative score adjustment process for grant applications using a blend of statistical, natural language processing (NLP), and machine learning approaches. The score adjustment process significantly mitigated bias incurred by expert judge reviews, increasing the odds of surfacing highly impactful grant proposals.

Streamlined philanthropic prize competitions by machine learning approaches to accelerate and de-bias the grant application review process.

Built data processing tools to help unlock the social impact of $1 billion in annual prize money awarded by Lever for Change

Relevant GitHub repos: Grant Proposal Clustering Model Pipeline, Grant Proposal Landscape Application, Grant Competition Scoring Pipeline

Designed & built web applications for a variety of clients. See portfolio for a detailed project list

Delivered secure & scalable solutions at minimal cost using best-practices in database architecting, data engineering & ETL, REST API, cloud computing services, and server-side programming

Researched and presented subnational ethno-religious conflict in Myanmar to U.S. policymakers using open source research and data visualization techniques

Represented the bureau at Departmental, think tank, and NGO meetings and shared main takeaways with colleagues

Researched evolving great power relations in the Asia-Pacific region to help policymakers construct new strategies for great power competition