Jeff Lyons Email & Phone Number

Currently developing innovative machine learning algorithms to replace current weapon planner software in the Integrated Air and Missile Defense System- Developed scenarios and generated high-fidelity Monte Carlo missile performance data- Conducted comprehensive data analysis to determine and derive critical features- Designed classification and regression models to predict probability of hit-- Performed hyperparameter selection using cross-validated grid searching with different scoring functions-- Assessed model performance using various performance metrics, learning curves, and validation curves-- Mitigated imbalanced data issues using minority class scoring metrics and sampling techniquesDesigned and developed analysis software in Python/Qt. This tool enables the ability to conduct simultaneous analysis of tactical telemetry data, digital simulation data, and hardware-in-the-loop data. Support for complex analytical queries from large Monte Carlo datasets facilitate multi-venue statistical analyses and performance insights that were previously impossible. Determined pre-flight test probability of success and assess potential flight test risks through simulation analysis. Investigated anomalous behavior and determined the root cause of each failure. Identified software bugs through code inspection of the C/C++, FORTRAN, and Ada. Verified algorithmic correctness of critical subsystems such as guidance, autopilot, target state estimation, and seeker signal processing by deriving models from first principles and comparing to software implementation.

Redstone Arsenal, Al

Designed and led development of an innovative distributed simulation framework to solve the complex problem of concurrently executing high-fidelity missile defense simulations. The framework provided a set of services supporting distributed time advancement, message passing, and target modeling.Designed and led development of the next generation high-fidelity digital simulation of the PATRIOT weapon system. Divested the legacy one-on-one FORTRAN simulation into separate multi-instance C/C++/FORTRAN mixed-language simulations built on top of the distributed simulation framework. Re-architected physics-based models to support the new framework time advancement, message passing, and target modeling capabilities. Performed exhaustive verification and validation of the simulations.Lead PATRIOT PAC-3 engineer to determine feasibility and performance of the PAC-3 missile using external radar data. Developed physics-based models in C++ and FORTRAN, integrated missile simulation with external radar simulations, and conducted analyses that resulted in the success of two live-fire flight tests. Discovered several critical mathematical and algorithmic errors that required tactical software updates prior to fielding.Integrated real-time missile simulation into an hwil lab, enabling high-fidelity missile performance assessments using actual system hardware and software. Developed bit-level message set necessary to support closed-loop missile flyout and implemented non-blocking Ethernet interface in the missile simulation to communicate with the facility. The performance assessments led to the early discovery of several system and software errors during program development.Developed a 6-DOF model of the PAC-3 interceptor in Matlab and Simulink. Modeled the airframe, actuator, autopilot, and midcourse guidance algorithms. Developed a three-loop autopilot and a non-linear autopilot to determine potential performance improvements against certain targets.

Master'S Degree, Electrical And Electronics Engineering

Thesis: Distributed Monitoring and Control and Physical System Modeling for a Laboratory Three-Tank System

Bachelor Of Engineering - Be, Computer Engineering