One account · reproducible · observable
Charles Trinh — Cloud and Infrastructure | Las Vegas, NV · Las Vegas, Nevada · Henderson · Clark County, Nevada · North Las Vegas, Nevada

Cloud and infrastructure work in a regulated setting is the discipline of building an estate that is small enough to understand, observable enough to operate, and evidenced enough to defend.
What this is
The default cloud estate grows by accretion. A service here, a bucket there, an account created for a project that ended. Within two years nobody can describe the whole thing, and an estate nobody can describe cannot be secured, costed or audited. The first piece of work is almost always an inventory rather than an architecture.
The second is account and boundary design. A single account with good tagging is simpler to operate than twenty accounts with inconsistent controls, and simplicity is a security property. Where separation is required, it should be for a reason that produces a real boundary rather than for tidiness.
The third is observability. An estate that reports uptime but not error rate, queue depth or saturation is an estate that is monitored rather than observed. The difference shows up the first time something degrades slowly rather than failing outright.
The fourth is cost. In a regulated estate, cost is a proxy for waste and waste is often a proxy for unmanaged resources. A monthly review of what is running and why is a control as much as a budget exercise.
Finally, infrastructure is a compliance surface. Every control the organisation claims has an infrastructure expression: encryption, logging, backup, access. If the infrastructure cannot produce the evidence, the control does not exist as far as an examiner is concerned.
Identity is the control that everything else depends on. A cloud estate with a single identity provider, enforced multi-factor authentication and role-based access is dramatically easier to defend than one with local accounts scattered across services. Consolidating identity is usually the highest-value infrastructure change available, and it is also the one that removes the most operational friction.
Backup and restore are frequently confused. A backup that has never been restored is a hypothesis. The control that matters is the restore test, with a documented result and a measured time, because that is what determines whether an outage becomes an incident or a disaster.
Network design in a regulated estate is about paths rather than zones. 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. Drawing the paths explicitly, rather than assuming the diagram implies them, is what makes the boundary real.
Finally, the estate has a lifecycle. Services are deprecated, runtimes reach end of life, certificates expire. A register of what is running, at which version, with which support date, is the artefact that prevents an outage caused by something nobody remembered was there.
There is also a documentation question that determines whether any of this survives. An estate described only in the heads of the people who built it is an estate that degrades the moment they leave. The inventory, the boundary design, the path diagram and the lifecycle register are the four documents that make the estate transferable, and they are cheap to maintain once they exist.
The final consideration is the relationship between infrastructure and the application teams that use it. An estate that is designed without their input becomes something they work around, and the workarounds are where the unmanaged resources come from. Involving the teams that deploy into the estate in the boundary design is what makes the boundary hold, because a boundary people understand is a boundary people respect.
A last point on scale: the estate should be sized to the organisation that operates it. A design that requires a platform team the organisation does not have is a design that will be operated badly. Choosing services for the control they provide, rather than for the abstraction they offer, is what keeps an estate operable by the people who actually run it.
How I do it
I begin with an inventory of accounts, services, data stores and network paths, reconciled against billing so that nothing is missed. That reconciliation is the step that turns a guess into a picture.
From there I design the boundary: which accounts exist, what each is for, and what controls are applied uniformly. Tagging is enforced rather than encouraged, because a tag that is optional is a tag that is absent.
Observability is built around the four signals that matter: throughput, errors, latency and saturation, with alerts routed to whoever can act. Dashboards are built for the questions people actually ask rather than for the metrics that are easy to collect.
Cost review is scheduled monthly with a named owner, and resources that cannot be justified are removed rather than left running.
Finally, each control is mapped to its infrastructure evidence: the configuration export, the log query, the backup restore result. That mapping is what makes the estate defensible.
Identity is consolidated first: one provider, enforced multi-factor authentication, role-based access, and no local accounts except where a documented exception exists.
Restore tests are scheduled and measured, with the result recorded. A backup without a restore test is treated as unverified rather than as a control.
Network paths are drawn explicitly, with each path documented, restricted and logged. The diagram is reviewed against the actual configuration rather than the other way round.
A lifecycle register records every service, runtime and certificate with its version and support date, reviewed on a schedule.
The four documents are maintained as part of the change process rather than as a separate exercise, so they stay current because updating them is part of shipping rather than a task someone remembers to do.
Application teams are involved in the boundary design and in the tagging standard, so the rules are ones they can follow rather than ones imposed on them. That involvement is the difference between a standard that is adopted and one that is evaded.
The handover is the four documents plus the control-to-evidence map, reviewed with the team that will operate the estate. A handover that ends with a document nobody has read is not a handover, and the review is what makes it one.
Cost and control are reviewed together, because the resources that are hardest to justify are usually the ones nobody is watching. A monthly review that names an owner per resource is a control as much as a budget exercise.
Artifacts you keep
- An inventory of accounts, services, data stores and network paths reconciled to billing
- A boundary design with uniform controls and enforced tagging
- Observability across throughput, errors, latency and saturation with routed alerts
- A monthly cost review with a named owner
- A control-to-evidence map for every claimed safeguard
Proof
- 67 public hostnames, 51 serving, one AWS account
- ONC (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — public suite URL
- 104 clearinghouse enrolments across 53 jurisdictions
- 67 public hostnames, 51 serving, one AWS account
- SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564
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
The estate is a document, not a memory
A regulated estate that exists only in the heads of the people who built it is not an estate; it is a set of habits. The first property that matters is rebuildability: take any host, describe it from what is on disk, rebuild it, and know within minutes whether it came back correctly.
That property is not about tooling. It is about whether the description exists. Configuration management, container images, or even a careful runbook all satisfy it; what does not is a console history and one person who remembers the order of operations.
The practical test is an outage simulation: pick the host nobody wants to lose, destroy it in a staging environment, and rebuild from the repository. The gaps that appear in that exercise are the ones that would have appeared at three in the morning during a real failure.
One account, environments separated by policy
Account-per-environment is a legitimate pattern and it is not the only one. What matters is that the boundary between production and everything else is enforced rather than described: separate credentials, separate networks, separate data, and a policy that makes a production write from a staging principal impossible rather than discouraged.
A single account with that policy is cheaper to operate and easier to audit, because the controls live in one place. It fails when the policy is aspirational — when the same role can reach both — and then the separation is an illusion with a diagram behind it.
The check is simple: attempt the forbidden action with the staging credential and confirm refusal. If it succeeds, the environments share a blast radius regardless of what the naming convention says.
Backups, restores and the drill
Backups are files. Recovery is a capability. The only evidence that distinguishes them is a drill with a written result: what was restored, into what, how long it took, and whether the service worked afterwards.
Immutability matters more than frequency. A backup that can be deleted by the credentials used to run production is not protection against the most likely attack on a regulated estate, and the retention policy that keeps thirty daily copies is doing less than the one that keeps an unchangeable monthly.
Drill quarterly, with the call included. Recovery at three in the morning is a communications exercise as much as a technical one, and the person who needs to be reached should have been reached once before, deliberately.
Secrets, certificates and the quiet expiries
Two classes of outage account for most of the downtime on a well-run small estate: a secret rotated in one place and not another, and a certificate that expired while every internal dashboard showed green.
Secrets belong in one store, read at run time, with an owner and a rotation date. Certificates belong to automation with an external check, because the check that runs from inside the estate tests the estate's own opinion of itself.
Both are boring to fix and expensive to leave, and both are discovered by an external monitor or by a user rather than by anything the team already watches.
What good looks like in a month
By the end of the first month you should be able to produce four things on demand: the hostname inventory with owners, the restore drill record with a date, the secret inventory with rotation dates, and the list of what would break if the largest single dependency disappeared.
An organisation that can produce those four has an estate. One that cannot has a collection of running services and a belief about how they fit together.
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
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
- Phone: 702-559-3489
- Email: charles@trinh.io
- Based in Las Vegas, Nevada · working remotely across the United States
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 — cloud and infrastructure — 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.
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.