Artificial Intelligence/Reinforcement Learning Engineer

Torch Technologies, Inc.•Redstone Arsenal, AL

3d•Onsite

About The Position

Torch Technologies is seeking an Artificial Intelligence/Reinforcement Learning Engineer to research, develop, implement, and evaluate reinforcement learning and machine learning solutions for defense modeling, simulation, and decision-support applications. The position applies reinforcement learning to complex decision-making problems, builds simulation-based training environments, designs reward functions and observation/action representations, and evaluates algorithm performance. The engineer will work with software engineers, simulation developers, and subject matter experts to develop AI/ML solutions for military applications. The role includes implementing and training reinforcement learning agents in Python with Ray RLlib; prototyping and analyzing algorithms in MATLAB; integrating agents with simulation systems; and creating reproducible training and evaluation workflows. The engineer will conduct experiments, use distributed computing and GPU resources to improve training efficiency, analyze and document findings, and help transition successful research concepts into operationally relevant software prototypes. The position emphasizes hands-on research, software engineering, experimentation, and algorithm development rather than personnel management. The role is based at a customer site, follows an M–F, 8–5 schedule, and may require travel up to 10%. Relocation assistance is not available. The position is not contingent upon an award of contract.

Requirements

US citizenship.
Bachelor's degree in Computer Science, Computer Engineering, Software Engineering, Mathematics, or a related technical discipline.
At least 3 years of relevant professional experience in software engineering, machine learning, reinforcement learning, or numerical algorithm development.
Strong Python programming skills, including experience with scientific computing and AI/ML software.
Experience developing or evaluating reinforcement learning algorithms using Ray RLlib.
Experience using MATLAB for algorithm development, data analysis, simulation, or mathematical modeling.
Understanding of fundamental reinforcement learning concepts, including policies, rewards, observations, actions, and agent-environment interactions.
Experience with machine learning frameworks such as PyTorch, TensorFlow, or comparable libraries.
Strong analytical, debugging, and problem-solving skills.
Ability to communicate technical findings through written documentation and engineering discussions.

Nice To Haves

Master's degree or PhD in Computer Science, Artificial Intelligence, Mathematics, or a related field.
Experience with continuous observation and action spaces and algorithms such as PPO and SAC.
Experience with Gymnasium, Ray Tune, or other AI/ML experimentation and optimization frameworks.
Experience developing custom reinforcement learning environments.
Experience with defense modeling and simulation tools such as AFSIM or OneSAF.
Experience with time-series analysis, optimization, or numerical computing using MATLAB.
Familiarity with C++ and Python/C++ interoperability.
Experience using Linux, GPU-accelerated training, and distributed computing.
Experience with Singularity/Apptainer or other container technologies.
Experience conducting applied research for military, aerospace, or defense applications.
Familiarity with battle management, sensor management, engagement planning, or tactical decision-support problems.

Responsibilities

Research, develop, and evaluate reinforcement learning algorithms for complex decision-making and optimization problems.
Implement and train reinforcement learning agents using Python and Ray RLlib.
Develop and evaluate algorithms including PPO, SAC, and other applicable reinforcement learning approaches.
Design observation spaces, action spaces, reward functions, and episode termination conditions.
Evaluate training stability, convergence, generalization, and algorithm performance under varying environmental conditions.
Perform hyperparameter tuning, comparative algorithm evaluations, and controlled experimentation.
Analyze experimental results and recommend improvements to model architectures, training configurations, and reward strategies.
Develop and maintain Python-based reinforcement learning software, numerical algorithms, and experimentation utilities.
Use MATLAB for mathematical analysis, algorithm prototyping, performance evaluation, and visualization of experimental results.
Translate research concepts, mathematical models, and analytical prototypes into maintainable software implementations.
Develop utilities for processing simulation data, analyzing performance metrics, and visualizing results.
Apply software engineering practices including code reviews, testing, version control, and technical documentation.
Develop and integrate reinforcement learning environments using Gymnasium or comparable frameworks.
Connect reinforcement learning agents with modeling and simulation systems for training and evaluation.
Support integration between Python-based AI/ML applications and simulation software developed in C++ or other languages.
Develop reproducible training and evaluation workflows for simulation-based experiments.
Collaborate with simulation engineers and domain experts to ensure environment models represent research objectives.
Develop configurable training pipelines that support repeatable experiments.
Use distributed computing and GPU resources to improve training efficiency.
Collect and analyze performance metrics from training and evaluation experiments.
Document experimental methods, results, limitations, and technical findings.
Support transition of successful research concepts into operationally relevant software prototypes.