aws · baa · key custody

Charles Trinh — AWS Healthcare Architecture

Charles Trinh — AWS Healthcare Architecture
Charles Trinh — AWS Healthcare Architecture

Eligibility is a condition

The agreement covers eligible services used in eligible ways, and the second condition is the one that quietly fails when a managed service is adopted for convenience.

The defensible position is a service table with an eligibility reference and a check date on every row.

Boundaries and keys

A private subnet is not a control on its own. What matters is which paths exist into it, who holds credentials for them, and whether every path is logged.

Encryption at rest is table stakes; key custody decides who can decrypt, which decides who the data is entrusted to.

Operability

Restore drills quarterly with time to recovery recorded, because that number is the only one that matters during an incident.

Synthetic monitoring from outside the estate, since internal dashboards show green while a certificate has been expired for an hour.

What I run

A 51-hostname estate on a single AWS account under an active business associate agreement, with a 337-table clinical store and eight X12 005010 transaction sets.

Cost-controlled by choosing services for the control they provide rather than the abstraction they offer.

The eligible services question

Only a subset of AWS services may handle PHI under the business associate agreement, and the subset changes with the agreement rather than with the popularity of the service. A service adopted because it solved a problem cleanly can place PHI in a context the agreement does not describe, and nothing breaks, which is why the drift is discovered at review rather than at adoption.

Keep a table: service, eligibility reference, date checked, person who checked. It takes a day to build the first time and ten minutes to review each quarter, and it is the artefact an examiner reads when the question is which services hold PHI and how you know.

One account, honestly separated

A single AWS account is cheaper to operate and easier to audit when the boundary between production and everything else is enforced by policy rather than described by naming. It fails when the same role can reach both, and then the separation is an illusion with a diagram behind it.

Test the boundary by attempting the forbidden action with the non-production credential. If it succeeds, the blast radius is the account regardless of how the environments are labelled, and the finding is the policy rather than the naming.

Key custody is the real decision

Encryption at rest is table stakes and every regulated estate claims it. The decision underneath is custody: whoever can decrypt is whoever the data is ultimately entrusted to, and a customer-managed key stored in the same account as the data is a boundary in name only.

Where keys are managed internally, the rotation, revocation and recovery paths need testing rather than documenting, because a key ceremony that has never been performed is a plan. Test the recovery once and record the time, since that number is what an incident will ask for.

Logging that answers a question

CloudTrail, flow logs and access logs exist in almost every regulated estate, and the question is whether anyone has asked a specific question of them and received a specific answer within the hour. Who deleted this record, what changed on that policy, which host spoke to that endpoint.

Run those three queries as a rehearsal before an incident. The gaps that appear are usually retention, a missing field or a service whose logs go somewhere nobody federates, and each is cheap to close while the question is hypothetical.

Cost control without losing control

A regulated estate runs lean when services are chosen for the control they provide rather than the abstraction they offer. Managed convenience that removes an audit path is expensive in the currency that matters, and the bill does not show it.

Review cost per hostname and per workload annually against growth. The review produces two useful questions: what could be consolidated, and what is growing faster than the business. Both are cheap when asked early.

Network boundary in practice

A private subnet is a set of routes, not a security control. What matters is which paths exist into it, who holds credentials for each, and whether every one of them is logged. Enumerate the paths and the credentials, and test the ones you did not expect to find.

Where a path cannot be closed, record it with an owner and a compensating control. The register of paths is the artefact that answers the question during an incident about how somebody reached a system they should not have.

Backups and the drill

Backups are files; recovery is a capability. The only evidence that distinguishes them is a restore into a clean environment with a written result: what was restored, how long it took and whether the service worked afterwards.

Immutability matters more than frequency for the most likely attack. A backup deletable with production credentials is not protection, and the retention policy that keeps thirty daily copies is doing less than the one that keeps one immutable monthly.

Certificates and the quiet expiry

The outage that takes a surface down without an application error is almost always a certificate. Monitor expiry from outside the estate, automate renewal, and verify the automation by letting it renew something rather than by reading the configuration.

Keep the inventory of hostnames and expiries with an owner, because the certificate nobody tracked is the one that expires during a holiday.

What to hand over

Four things: the service table with eligibility references, the path register, the restore record and the certificate inventory. Together they describe an estate somebody else can operate, which is the standard for a handover.

Anything beyond those four is depth for a specific incident, and depth is valuable and not a substitute for the four.

How an engagement runs

A week of tracing one workload from the public edge to the record store, with a configuration export at each hop and a note of who holds credentials for it. The output is a path register and a list of the controls that exist only in the diagram.

Then the evidence pack: the service table, the restore record, the certificate inventory and the three log queries that answer the questions an incident would ask. Each item is dated and each has an owner.

The engagement ends when somebody else can operate the estate from those four artefacts, which is the only handover standard that survives a change of staff.

Where the analysis stops being technical

The hardest questions in a regulated estate are not about services or encryption; they are about who is allowed to say yes. A change that touches PHI needs an approver who understands the obligation, and most organisations have that person only in principle.

Naming them, and naming a deputy, costs an afternoon and removes the largest single cause of delay in a change process that is otherwise well designed.

Proof

  • ONC (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — public suite URL
  • 67 public hostnames, 51 serving, one AWS account
  • 104 clearinghouse enrolments across 53 jurisdictions
  • USCDI v7 proposal — Patient Digital Identity Credential — listed at isp.healthit.gov
  • SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564
  • AWS Business Associate Addendum active since 18 August 2026

Where this work is delivered

Engagements run from Las Vegas and reach across the valley: clients in Las Vegas, NV; Las Vegas, Nevada; Henderson; Clark County, Nevada; and North Las Vegas, Nevada — plus remote delivery nationwide.

How an engagement runs

1. Trace one real case

Follow a single patient or claim end to end with a timestamp at every hop, before any document is written.

2. Written findings

A findings list naming the defect, the system, the owner and the date.

3. Monthly review

Review the list monthly and check whether the dates held.

Frequently asked

What is this service?

This page describes charles trinh — aws healthcare architecture — the scope, the failure modes it prevents and how the work is measured. It sits in the Cloud and Operations category at trinh.io.

Where is the work delivered?

Engagements run from Las Vegas, Nevada across the valley — Las Vegas, Henderson, North Las Vegas, Spring Valley, Paradise, Winchester, Sunrise Manor, Enterprise, Summerlin, Clark County — and remotely anywhere in the United States. On-site work in the Las Vegas valley, remote delivery elsewhere.

What does it cost?

Standard rate $195 per hour, committed $165 per hour at twenty hours or more per week for three months or more; $1,600 standard day, $1,350 committed day. Scoped in writing before work starts, fixed fee where the deliverable is defined.

Who is Charles Trinh?

A technology architect in Las Vegas, Nevada, working in healthcare identity, consent and interoperability. Not a physician, not a medical practice, and not affiliated with any other Charles Trinh.

How does an engagement start?

With one trace of a real case and a written findings list naming the defect, the system, the owner and the date. The first call is twenty minutes and free: 702-559-3489.

Engagement

Scoped in writing before work starts; fixed fee where the deliverable is defined. First call is 20 minutes and free. 702-559-3489 · charles@trinh.io

Talk to Charles

Record

Profile

Healthcare technology architect working where federal regulation becomes production infrastructure. Built and operates a FHIR R4 clinical platform — SMART App Launch authorization server, patient and provider portals, SQLite — on a hardened single-host AWS estate of 67 public hostnames. The platform returned a self-test PASS on the ONC (g)(10) Standardized API test suite in April 2026, 317 of 317 subtests, publicly verifiable. Designed a consent architecture in which a permissioned ledger records patient grant, refusal and revocation as immutable events while all clinical data stays off-chain and encrypted per patient — no PHI on the ledger by design. Filed a proposed USCDI v7 data element with ONC on verifiable patient identity. Translates the Federal Register, HHS/CMS guidance and HITRUST requirements into architecture, controls and audit evidence.

Selected engineering scale

  • Self-managed estate of 67 public hostnames, 51 serving, on one AWS account.
  • FHIR R4 API with a SMART App Launch authorization server; patient and provider portals; an AI service layer; Hyperledger Fabric peer, orderer and CA; IPFS; a 337-table SQLite clinical store.
  • Two permissioned distributed ledgers in production, one operating a consent registry that records grant, refusal and revocation with no PHI on chain.
  • X12 005010 pipelines across eight transaction sets.
  • Designed, built, deployed and operated end to end by one person.

Credentials and standards work

  • Inferno (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — the ONC Standardized API test suite. Standards validated: US Core 6.1.0, USCDI v3, SMART App Launch 2.0.0, Bulk Data 2.0.0. Public suite URL.
  • USCDI v7 public comment submission, April 2026 — proposed Patient Digital Identity Credential under the Provenance data class, listed publicly on the ONC Interoperability Standards Platform.
  • SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564. AWS Business Associate Addendum active 18 August 2026.

Technical environment

AWS us-east-1 (compute, storage, messaging, SMS, email, voice) · Linux and systemd hardening · nginx · Fabric and EVM ledgers · FHIR R4 · X12 005010 · HL7 · SQLite · self-hosted mail, drive, search and monitoring · Git-based change control and CI release gates.

TRINH — Charles Trinh, technology architect
TRINH

Charles Trinh

Fractional CTO/CIO and Technology Architect — Healthcare Data, Identity & Regulated Infrastructure · Las Vegas, NV

Las Vegas, Nevada · US citizen · Fractional and project engagements

(702) 559-3489 Email Charles Trinh