Kochise Bennett

Background

About Kochise Bennett

Applied machine learning researcher specializing in deep learning with extensive experience in computer vision and reinforcement learning. I research state-of-the-art techniques and apply them to a variety of problems, often by designing custom algorithms or adapting best-practices as needed for the relevant problem. Particular applications include marine mammal detection, satellite image analysis, multi-agent RL for continuous control, and run-time assurance for safe autonomy. Extensive software development experience including custom-built supervised learning pipelines (tensorflow, pytorch, or jax) for image classification, object detection, and image segmentation tasks. Extensive software development experience designing simulations and training RL agents using ray.rllib and AFRL's CoRL package (https://github.com/act3-ace/CoRL). Research experience also includes neural net architecture development, explainability/interpretability tools, transfer learning and low-data scenarios, and time-series modeling.