Last entry. Capstone Festival is today, portfolio and Learning Journal
are submitted, and v1.0.0-capstone is tagged. CSUMB is done.
Looking Back
Reading back through two years of journals, the early entries read like someone who knew what they wanted but wasn’t sure they could pull it off. I came into this program with 10+ years of aerospace engineering behind me, but almost no formal software development foundation. What I learned across CST338, CST438, and beyond is that the engineering discipline I already had translated directly.
The biggest growth wasn’t technical — it was learning to be a beginner again. Walking into a Python-first, Git-first, cloud-first curriculum after a decade of MATLAB and Simulink required a real identity reset, and the early journals are full of that friction. The later ones read differently.
The Capstone
The capstone, ggSwarm, became the project where everything converged: a physics-based drone swarm simulation framework built in NVIDIA Isaac Lab, training a shared-policy PPO agent across 4,096 parallel GPU environments using GATv2 graph attention networks, MINCO trajectory optimization, and CBF safety constraints. It’s the most honest expression of what I was always trying to do: close the gap between what a model predicts and what a physical system actually does.
What’s Next
Going forward I’m pursuing Modeling & Simulation and Autonomy Simulation roles in defense tech, and continuing to build out my C++ simulation portfolio. The program gave me the foundation; the work from here is applied.