A start-to-finish walkthrough of the production platform: wallet connection via EIP-6963, SIWE authentication, on-chain token minting, session attestation on Sepolia, and the dashboard that ties it together. Every step runs against live contracts and a real backend — nothing mocked.
The session shown is an original bilateral-stimulation exercise designed and built from scratch: a canvas-rendered target oscillating left-right on a cosine-interpolated path — roughly 45 seconds, 20 passes, driven by requestAnimationFrame. Completing it produces a wallet-signed claim; the server verifies the signature before minting 1 HNT on Sepolia and records the session with its transaction hash, so the dashboard reads live on-chain state back to the participant — the full loop from interaction to verifiable record.
Verify: SovereignLedger v2 on Blockscout
HNT (FSLToken) · EHT v2 · HypnoNeuroNFT · SovereignLedger v2 · AlchemistForge · SovereignSession · NeuroBalanceConsent · SovereignAchievement · BenevolenceFund v3 · MerkleAnchor · BenevolenceFund v2 (superseded)
Wearable data, consent-gated — architecture and live verification
Most recently a consent layer for wearable health data — a Google Health API integration mapping readings to FHIR R4 Observations with LOINC coding, gated behind a participant-signed permission record with scoped, time-limited access. The gap analysis, architecture, and staged adoption roadmap are published alongside it.
A five-surface hybrid platform for health data where the patient holds their own credentials. Consent is captured as a cryptographic signature from the participant's wallet, attested on-chain, and enforced across web, session, and storage layers.
I own it end to end: the authorization model, the session and identity layer, the component system the surfaces are built from, and the smart contracts underneath. Eleven contracts deployed and source-verified on Ethereum Sepolia — ten current, one superseded — covering token issuance, consent governance, session attestation, and record anchoring.
Before this I spent thirty years in clinical practice. Traveled between practices in one of the Southeast's largest dental service organizations, standing up hygiene departments in offices that had no provider. Documented at Durham: per-provider production $380,710 to $624,640 annualized from ten months, a 64% increase across four documented years. Founded and ran a consumer health product line end to end, formulation through retail, then moved it direct-to-consumer in 2020.
The interface decisions came from thirty years of clinical practice, not a style guide. Where the consent step sits, how much a screen asks for at once, what someone needs to see before they sign — those came from watching people navigate care, then building it and testing it directly. The component system underneath is documented in Storybook with tokens, variants, and accessibility annotations.
I scope work with agent contracts: each task names the files it touches, the acceptance criteria, and a standing rule that nothing outside scope gets modified. This eliminates drive-by refactors and keeps diffs reviewable.
Every build artifact is verified against what it claims to be. Deployed contract addresses are checked against Blockscout on-chain state, not configuration files; served pages are curl-verified against the live domain, not the build output. Every public-facing assertion — “verified,” “deployed,” “live” — is traced to the artifact it describes. If the artifact contradicts the document, the document is wrong.