Technical dossier / Selected work
Four systems. One standard: show the evidence.
Case studies explain the problem, my role, the architecture, proof points, and what the current evidence does not establish.
01
Private pre-release product
A clean, local-first macOS voice workspace for dictation, meetings, evidence-grounded notes, and structured memory.
Deterministic vocabulary correction reduced delivered-text WER from 4.74% to 4.09% on Echo’s retained 13-fixture generated regression suite.
- Swift
- SwiftUI
- Core ML
- FluidAudio
- MLX Swift
02
Private business project
A model- and runtime-independent reference for permissions, mission contracts, artifact validation, and evidence-backed AI workflows.
The reference workflow blocks undeclared permissions and cannot pass until its required evidence and report artifacts satisfy explicit validators.
- Python
- Typed contracts
- Workflow gates
- Audit evidence
03
Open source
A brief-driven Python engine that plans, selects, and packages short-form ad campaigns into reviewable production artifacts.
One validated brief produces six concepts and three stable ad specifications plus posting copy, QA, and a manifest; production ranking remains private.
- Python
- YAML
- Ollama
- Provider adapters
- Pytest
04
Open-source tools and attributed adaptation
A public lead-research demonstration and an attributed cloud-agent runtime evaluation with private production policies.
The local lead agent can run from labeled fixtures, while the cloud system keeps outreach draft-only until explicit approval.
- Python
- Ollama
- Typed policy contracts
- TypeScript
- Vercel Sandbox