Slatron
Distributed real-time broadcast infrastructure with embedded AI. Rhai scripting, local ONNX-based audio inference, audio ducking, zero-downtime scheduling, multi-arch cross-compilation to a single self-contained binary.
PhD Candidate LSU Cybersecurity Lab
Studying quantum threat models and AI agent trust.
Building the orchestration layers that make autonomous software production real.
From photons to syntax.
Computers & Security, Volume 145, 2024. doi:10.1016/j.cose.2024.103997
American Journal of Physics, 91(9), 708–713, 2023. doi:10.1119/5.0105418
Q-NET Symposium 2025 — Poster. PDF
Distributed real-time broadcast infrastructure with embedded AI. Rhai scripting, local ONNX-based audio inference, audio ducking, zero-downtime scheduling, multi-arch cross-compilation to a single self-contained binary.
Rust-powered toolchain for C codebases: documentation generation, static analysis, IDE integration. Rethinking how low-level infrastructure is documented and maintained.
Local-first AI orchestration pipeline: on-device transcription, LLM-powered semantic analysis, encrypted storage. A study in private, sovereign AI infrastructure where the user controls data and models end-to-end.
Interactive 3D Bloch sphere visualization for quantum state pedagogy. Built to make quantum computing tangible — the kind of tool I wished I'd had when I started.
Single-user desktop orchestrator (Electron + TypeScript) that runs two lanes in parallel: PR review in a sandboxed worktree that physically cannot push or call gh, and software development through a full task lifecycle (claim → implement → draft PR → shepherd to ready-to-merge). Pluggable adapters for trackers (Jira, GitHub Projects, Vikunja, Azure DevOps Boards), forges (GitHub, Azure DevOps), and harnesses (Claude, Codex, Cursor, OpenCode, pi). The brain is a deterministic poll loop; LLMs are called only for judgment; PR approval and merge are always an explicit human click.
Rust implementation of the Policed Agency architecture: a capability-gated policy engine for AI agents, with multi-tier escalation (executor → guardian → human) and an asymmetric communication protocol that prevents the executor from tainting the safety layer. Treats agents like employees, not root users.
I write across five lenses — theory, physics, systems, agents, reflections. The breadth is intentional.
A formal security architecture for AI agent delegation: capability-gated IAM, multi-tier escalation, asymmetric trust protocols.
Notes from a security researcher who wants to believe — and the surface area that makes belief expensive.
The unsung hero of Shor's and Grover's, explained with the matrix algebra most resources skip.
A grounded reading list and a realistic timeline for beginners.
From a failed prototype to a robust distributed system in Rust — cross-compilation, embedded UI, Rhai scripting.
Embedding a local ONNX-based AI DJ into broadcast infrastructure.
On REPL-driven development, the order of operations, and first-class functions.
A post-mortem on a language's hype, and what its dual paradigms actually buy you.