alan chester.
∀ ε > 0, ∃ δ > 0Research focus
I work at the intersection of cloud infrastructure, agentic AI systems, reinforcement learning, and multi-agent trust. The site you are reading is where the research and technical work lives. Notes from the bench.
The active empirical work: in long-horizon reinforcement learning, do symbolic observations beat pixel observations once encoder capacity is properly controlled? Pokémon Red is the environment, RecurrentPPO via SB3-Contrib the method.
Current questions
On reinforcement learning.
How does observation representation affect sample efficiency in long-horizon RL?
Pixel vs. symbolic vs. hybrid observations under capacity-matched encoders.
On team architecture.
How does the unit of work change when teams include autonomous agents?
What structures support variable-elasticity teams.
On multi-agent security.
How does trust topology affect security in multi-agent LLM systems?
Centralized orchestrator vs. peer-authenticated trust.
Currently
- Studying observation representations in long-horizon RL. Pixel vs. symbolic vs. hybrid, under capacity-matched encoders.
- Researching team architectures and trust topology for human-AI systems.
- MBA candidate, NC State Jenkins Graduate School of Management.
Background
About fifteen years across kernel engineering, Kubernetes platform development, cloud security, and product leadership.
Dual B.S. in Computer Science and Applied Mathematics, NC State. CS gave me the discipline to ship systems that run in production. Applied Math gave me the internals of the models the rest of the field is now using. Security taught me to see the risk others miss. The MBA in progress is the habit of proving ROI rather than asserting it.
Philosophy
The brand is borrowed from real analysis. ε is whatever the world demands. Uptime, accuracy, security, ROI. δ is the move that meets it. Data points the gradient.
Connect