Software Design Engineer And Predictive Analysis (Machine Learning) Team Lead
CurrentU.S. Army Combat Capabilities Development Command (CCDC) Aviation & Missile Command U.S. Army's Air Missile Defense - Science and Technology (S&T) programsArmy Rapid Capabilities and Critical Technologies Office programResearched, designed and prototyped a distributed architected simulation tool suite for fire control designs and algorithms. Allows for experimenting, modeling and demonstrating complex air and missile defense systems with prospective fire control algorithms and design enhancements. Researched machine learning regression and neural network algorithms to develop the capability of predicting the divert capabilities of simulated missiles. Exploring the usage of RNN, LSTM, and GRU for time series prioritizing of threat shot patterns using directed energy weapons (DE M-SHORAD). Created classification service to predict classification of AFSIM-based simulated threats using machine learning algorithms.Developed and trained machine learning models using Python Sklearn, TensorFlow and Keras packages. Developed capabilities to config machine learning algorithms and hyper-parameters via YAML/JSON configuration files to expedite development and training of Python machine learning models. Developed approach to apply normalization and feature encoding for times series datasets.Helped with the creation of a user interface in React for investigating air defense simulated scenarios with Commander/Gunner firing controls. Lead research and prototyping of Microsoft HoloLens 2 and Unreal Engine 4 capabilities for creation of Augmented Reality overlays for military control stations. Leading machine learning team of engineers to build a reinforcement learning data processing pipeline around AFSIM simulator. Involving the integration of Python Gym RL framework with C++ services and AFSIM for action handling into AFSIM as well as observation and reward feedback to RayRL and Stable Baselines3 for RL policy training.