San Francisco Bay Area
Hi! I'm a Senior Software Systems Engineer at Zoox with a background in mechanical engineering and data science. I've got 10 years of experience developing digital twins of the real world, from jet engines to commercial real estate buildings. For three of those ten years, I've led a team responsible for implementation and delivery of a customer facing application and the core models that drive it. I have a passion for combining data science and physics based modeling to solve real world problems. Relevant Skills: Python (and its many packages [pandas, matplotlib, etc]), Neural Networks, Optimization, MATLAB, Monte Carlo Analysis, NPSS, Unigraphics, Solidworks contact [at] paulmorrison.io
- Leveraged modern LLM prompt engineering methods to build an AI powered utility bill parsing tool to enable a freemium product offering. - Developed a novel weather normalization algorithm to help clients understand the contribution of weather to differences in their utility bills. The algorithm is a refinement and extension of an artificial neural network building energy use model. - Designed and implemented an automated fault resolution system using data-driven heuristics, resulting in significant efficiency gains for internal engineers and proactive responses to clients recognizing their actions. Overall, client notification engagement open rates increased to over 50%. - Doubled deployment frequency through refactoring and streamlining the codebase, while seamlessly iterating toward a new client-facing application built on a simpler frontend technology. Upgraded key dependencies of the main application with no breaking changes and zero downtime.
- Supported technical and product roadmap development, coordinating resources and achieving an on-time delivery rate of over 90%, while championing a user-centric development approach, driving frequent releases and iterative improvements based on feedback. - Managed a team of model development and data science experts while personally contributing to key projects such as legacy product data migrations and prototyping core components for the fault detection and quantification algorithm. - Revamped our competency matrix to empower IC ownership of their professional development resulting in 3 promotions, 1 role transition into my software team, and successful onboarding of 3 new hires. - Reduced model implementation timelines from 6 weeks to less than one week through specification simplification for third-party developers and team process optimizations like elimination of unnecessary meetings and a blameless "andon-cord" system.
- Conducted deep dives into thermodynamic models and reduced model execution time by up to 45% by fine tuning iteration loops within legacy models. - Optimized data collection requirements which reduced analysis time by 20% per building while still meeting model accuracy expectations set by the commercial real estate market. - Worked across Software and Sales teams to set up a repeatable process for linking disparate data sources to gain insight into the most important drivers of building efficiency. This pipeline and process was immediately used to set Engineering-wide team strategy.
Engine Systems & Performance - Led trade studies to optimize embedded model sensor suite for accurate engine health prognostics - Supported on-board analytics with model projection uncertainty quantification - Drove the use of physics-based class code into new product models for robustness and accuracy - Introduced collaborative software development practices including Git version control and code review
Engine Systems & Performance - Developed a neural network health estimator in Python with greater accuracy than legacy regression based estimators - Developed Python based data analysis and visualization applications using Matplotlib and Bokeh - Implemented a distributed simulation engine to enable high fidelity model assessment - Conducted trade studies in support of foreign military sales campaigns - Communicated effectively to customers and bridged the gap between performance and the analytics tech community
Engine Systems & Performance - Used Gaussian Process techniques to quantify uncertainty in engine-to-data model match - Customer point of contact for aircraft/cycle model integration and performance evaluation - Real-time support of engine test facility during accelerated endurance assessment