Portfolio

Systems, research, and things that work.

A portfolio of difficult problems made inspectable—from agent architecture and cybersecurity research to the infrastructure beneath them.

Each case study follows the problem, constraints, approach, architecture, outcomes, and next questions.

4 selected projectsResearch · Software · Infrastructure

softwarebuilding
C2 — A Single-Human Company OS
An experimental operating system for a single-human company: a deterministic operational spine, governed agent teams, scoped memory, and a structured model of organizational reality in PostgreSQL.
- Slack front door and human approval
- Alli orchestration and Missions
- Domain teams: managers, specialists, validators
View
experimentcomplete
Transformer-augmented reinforcement learning in the Unity Food Collector environment
Can a transformer beat the MLP inside PPO and SAC? A three-way study placing the transformer on the actor, the critic, or both, with a CoordConv-ResNet encoder, the failure modes that shaped it, and held-out results across five agents.
- Unity ML-Agents Food Collector (5×40×40 grid sensor, hybrid action space)
- Gymnasium wrapper over the Unity binary
- PyTorch
View
toolingarchived
Claude Kit: domain templates that turn Claude Code into a specialist
A template system for Claude Code: a base layer of agents, skills, commands, hooks, and rules merged with a domain layer and scaffolded into a project's .claude folder.
- Claude Code
- Node.js scaffolding and validation scripts
- Shell hooks
View
researchcomplete
Deepfake detection across generations: MesoNet to ViT
Nine detectors from 18K-parameter MesoNet variants to fine-tuned Xception and ViT-B/16, benchmarked across GAN-era, diffusion, and held-out forgeries.
- PyTorch
- MesoNet / MesoInception4
- CBAM attention with Grad-CAM
View