DRAFT

Spring 2026 CS 4850 Senior Research Project:

Reinforcement Learning on Supply Chain Dynamic Policy Development

Project Overview

Given an anonymized dataset representing demand history for inventory items, implement and train a decision policy (or model) using one or more reinforcement learning algorithms.

Preferred prerequisite knowledge:

Python, Numpy, Linux, AI, machine learning.

Knowledge of supply chain concepts is helpful, but not required.

Skills/Knowledge that students will learn:

Python, Numpy, Linux, JAX, PyTorch, Keras, AI, machine learning, reinforcement learning, research, and writing.

Major Work, Milestones and Expected Outcome

Major work and expected outcomes:

• Code that implements an offline reinforcement learner for the given supply chain problem
• Code that collects performance metrics
• Analyses of performance metrics
• Research paper suitable for submission

Milestones:

• Week 3: Skeleton RL agent that logs experience tuples from the dataset
• Week 6: RL agent(s) that learns from the dataset
• Week 9: Performance analyses of RL agent and comparison to quantum baseline, actual test set
• Week 12: First draft of report
• Week 15: Final report and code deliverables

Note: Week 15 is final exam week.

Ideas For Final Systems / Deliverables To Develop:

• Reinforcement-learning agent(s) that takes supply-chain decisions (buy, repair, etc.)
• Performance analyses
• Submission-quality research paper

Recommended Team Size

Two

Maximum Number of Teams Accepted For This Topic

One

References/Tutorials For Project

TBD

Project Publication and Procedure

1. Is this request in support of a research project? If yes, will the students get credit when the research is published?

Yes, and you will be an author! I will help you write the paper and, most importantly, teach you how to write it.

2. Is a Non-Disclosure Agreement (NDA) required?

TBD

3. Are you available to meet with the students on a regular basis (schedule to be determined by you and the team) and provide feedback on student progress?

Yes, and regular meetings will be required. Ideally you will join my research lab's weekly meeting. My ulterior motive is to identify strong candidates for the PhD program.

4. Do you want the student team to sign up for your project or would you prefer to interview them and select the team?

I prefer to interview candidates.

5. May students apply to, and present this project to, C-Day at the end of the semester?

Yes! I will help you with the poster and demonstration, if applicable.

Contact Information

Christopher.Simpkins@kennesaw.edu