I've gathered recorded screenshots, source evidence and test results in one place. I also note which projects are examples and which source repositories aren't public.
Detailed featured work with screenshots and architecture
The archived screenshots document earlier synthetic tests, alongside recorded outcomes and static diagrams showing the trust boundaries.
I put these small test projects together with AI assistance to study replay, retries, and token rotation. The tests ran locally; the GitHub repos are currently private. They aren't customer deployments.
TypeScript / Node.js crypto / node:test
ZombieMutex Replay Guard
One-time challenge, request binding, replay rejection and concurrency edge cases.
5 passing Node tests at publication
Not platform attestation; a distributed deployment needs an atomic shared replay store.
Download the TypeScript models, runnable Node.js tests and threat models for local inspection. The bundle includes the Break It / Fix It API cases and passkey/TLS/API decision checks. All data is synthetic, not private project or customer data.
These models do not prove cryptographic attestation, production authentication, operational telemetry, or a particular customer deployment.
Case-study evidence status
These are write-ups of systems I worked on and built in enterprise environments. The architecture diagrams are simplified, and customer source code, infrastructure details and internal results remain private.
Hardening iOS Device Trust Signals
I hardened iOS device-trust decisions by separating untrusted client signals from evidence verified by backend services.
Private implementation · public architecture notes
I worked on session-recovery controls that made identity checks, step-up decisions and failure states explicit instead of weakening authentication during recovery.
Private implementation · public architecture notes
I run TypeScript, lint, regression tests and a full site build before deploying. I also check the actual pages in a browser. GitHub Actions is configured, but GitHub currently blocks those runners because of an account billing limit.
The downloadable source bundle includes local tests and reference fixtures; it does not provide a hosted simulator.