Founder & software engineer.
AI systems. Security. Commercial software.
I’m building SellCode.ai, a new home for code licensing. My work spans AI security, data infrastructure, and robotics simulation. Based in San Francisco.
Founder · September 2026–Present
CURRENT VENTURE
I founded SellCode.ai to explore new ways for developers to monetize their code. The coming-soon site is live, with a waitlist for launch updates.
Four projects · actual application captures
A closer look at the systems I’m building. Open a capture to see the full interface, or inspect the demo notes below.
01 / AI security coursework
Inspect three defensive prompts, record expectations, and verify SQLite parameter binding locally. Built for my Open Avenues coursework.
In progress · offline foundation
The Assignment 1 interface has three examples, prompt inspection, and support for downloading an experiment plan. A local SQLite check verifies the binding example. Live Gemini responses are still pending; this is the foundation for a security agent.
02 / Evidence and diligence
Choose a facility, inspect its sources and fuel mix, and export an evidence dossier from structured energy records.
Historical data · decision on hold
The Carson refinery capture uses 2014 records. Three sample facility dossiers support Markdown and JSON exports. Each remains on HOLD because the historical records do not establish current decision evidence.
03 / Robotics simulation
Build a quadruped, test movement controllers, and replay a trial through a 3D interface backed by Rust physics.
3D simulation · software trials
The capture shows a completed flat-ground trial with movement and stability measurements. Body settings and controller choices can be compared through repeatable simulations and replays. The parameters have not been calibrated against physical hardware.
04 / Causal security analysis
Trace an evidence chain and compare containment choices in a security investigation workbench.
Fictional scenario · modeled outcomes
The bundled Northstar scenario models an OIDC containment choice and projects reachable paths from seven to two. The comparison illustrates a local model’s outcome; it makes no changes to production systems.
Captured October 5, 2026.
Robotics evolution simulator
Axiom explores how robot body plans and neural controllers evolve together. Its Rust simulation keeps a diverse archive of candidates, making movement, fitness, and search history visible rather than reducing every experiment to a single winner.
Morphology genomes and neural controllers connect the shape of a simulated body with the way it moves.
A MAP-Elites archive retains different high-performing candidates across terrain tasks.
Seeded runs, checkpoints, and replay interfaces make it possible to revisit an experiment and compare outcomes.
From research and prototyping to tested, maintainable software.
LLM guardrails, retrieval workflows, NLP post-processing, and structured data pipelines. Python, PyTorch, DuckDB, and FastAPI.
Client penetration testing, prioritized remediation, AI governance, and zero-trust architecture. CompTIA Security+ certified.
AI-native developer tools and security software, supported by automated CI, tagged releases, and verification pipelines.
Robotics evolution, real-time telemetry, and interfaces that make complex systems understandable. Rust, C++, TypeScript, and React.
My commercial work includes private, licensed software. The highlights here share the problems I work on and the systems I build.
Software engineering, applied AI, security, and commercial software.
Independent Software Licensing
CyberTrust America
Popaisle
CCSF Tutor Squad
AI Tinkerers
CCSF Engineering Club
Computer Science · GPA 4.0
Computer Science
Deep Learning & Machine Learning Specializations; PyTorch for Deep Learning; Mathematics for ML & Data Science.
AI Security Agents coursework: Python and Gemini API experiments, defensive prompt evaluation, and local verification of security claims.