Bill Deans

Background

About Bill Deans

Working at SRI has been like working at a new company every few years. SRI performs research and technical services in many fields—communications, radar, remote sensing, molecular physics, chemistry, education policy, machine learning, theorem proving, natural language processing, and more—and every project needs software. Some of the projects I have worked on at SRI over the years: 2022-2025, Next-generation radar controller and user interface Designed and implemented the system controller and user interface for a next-generation software-defined radar, in collaboration with radar experts, electrical engineers, and system engineers. I wrote the system controller in Python, and the UI as an Electron application. 2016-2024, Active data assistant Worked with scientists and engineers at Pacific Northwest National Labs and customers to create a web-based assistant to support data scientists’ workflows, inferring high-level tasks from the user's actions and then recommending new data sources and analytic tools based on those high-level tasks. 2016-2017, Multichannel acoustic processor Worked with hardware and embedded systems engineers to design and build an embedded real-time beam-forming processor for acoustic data, which used the sample delay differences from multiple microphones to locate and enhance a particular sound source and reduce noise from other sources. 2009-2011, Automated air sample handling and analysis system Designed a distributed multi-platform (Windows and Linux), multi-language (C/C++ and Java) software architecture to automatically transport and analyze air samples at a rate of 100 per 24 hours, roughly a improvement in efficiency. One of my key contributions was the communications framework, which enabled us to orchestrate and monitor a myriad of analytic chemistry instruments and sample-handling apparatus from different vendors. 2007-2008, GPGPU for accelerated radar processing Prototyped a GPU-based synthetic aperture radar (SAR) processor to enable real-time imaging. I used OpenGL (CUDA was still in its infancy), and wrote a fragment shader to perform the compute-intensive back-projection and summing stages of SAR processing on an off-the-shelf NVidia “Gaming” graphics processor. The resulting demo was over times faster than the previous version, effectively making our radar real-time. This dramatically increased the utility and versatility of our radar system for customers, and was a foundational capability in SRI’s Desert Owl II program, a multi-year collaboration with US Central Command.