Lucas Doyle

ML Infrastructure

Led edge infrastructure / end-to-end implementation and deployment of video pipelines for Samsara's video-based safety products. These run on 2M+ of Samsara's AI dashcams, improving driver safety, engagement, and trust at scale.

• Developed and deployed pipelines (e.g. tailgating, rolling stop sign detection, lane departure), including device farm QA automation, shadow testing, and firmware infrastructure / feature development.
• Built internal tools for debugging, evaluation, and telemetry replay; enabled scalable model iteration and observability across firmware and cloud.
• Spearheaded offline evaluation pipelines for quantized models and integrated support for multimodal data (e.g. video, IMU, GPS, hardware emulation).
• Mentored interns and collaborated cross-functionally with product, firmware, science and data teams to streamline ML deployment and validation processes.

Mobile

• Led performance and infrastructure efforts for mobile applications with 500k+ DAU, driving major improvements in responsiveness, reliability, and developer velocity.
• Helped bootstrap new apps, products and the company’s first native Android systems and mobile device management (MDM) stack.

Developed core technologies for a drone platform to capture, upload and process large-scale imagery into survey-grade data for insurance and mining industries.

• Created a desktop app for flashing and configuring fixed-wing and multirotor drone avionics hardware.

• Built cloud workflows for large-scale survey processing and analysis with a GIS/geospatial-focused UI.

• Developed a mobile app for drone operators using React Native and iOS, enabling geofence editing and on-device photogrammetry processing.

• Led feature development, large-scale refactors, product mergers / acquisitions; contributed to hiring, planning, and mentoring engineers.

Fourth employee at a Japanese industrial robotics startup:

• Designed and implemented the web interface for the MUJIN Controller, a robotic arm workcell planning system used by Canon, Honda, and other system integrators.
• Developed customer-facing UIs featuring a real-time WebSocket-based WebGL viewer for a bin-picking system.
• Collaborated with an international team while living in Japan for two years.