Customer stories / Helvetia Health

How Helvetia replaced a
four-person platform team
with two engineers and Scalable.

Eighteen months ago, Helvetia ran 240 services on a self-hosted Kubernetes stack — three on-call rotations, a six-engineer-week audit cycle, and a CHF 1.9M annual infra bill. Today, two SREs operate the same workload from a single dashboard. The platform team became a product team.

Julia MarchettiVP Engineering · Helvetia Health
Industry · Healthtech Region · Switzerland · EU-pinned Employees · 2,400 Services · 240 On Scalable since · Aug 2024
Platform headcount4 → 2Two SREs replaced a full platform group. Two engineers moved to product.
Audit cycle40minLast EU residency audit. Down from six engineer-weeks.
Uptime (12 mo)99.997%Three nines plus a bit. No customer-reported incidents.
Infra bill−68%From CHF 1.9M / yr to CHF 612k / yr — same workload.

01The before

By the start of 2024, Helvetia's engineering org had grown to 84 people. Forty of them built the product. Twenty-two of them — between platform, SRE, and DevSecOps — kept production running.

The stack was the kind of thing you build when you have to. Self-hosted k8s on bare-metal in two Swiss data centers, ArgoCD, Vault, Istio, Loki, Tempo, a homegrown deploy CLI, and a 40-page runbook that began with the words "if you are reading this on a weekend, page Pavel."

The bill: CHF 1.9M / year in DC fees, licenses, and observability tools. The cost the bill didn't capture: an audit cycle that consumed six engineer-weeks every quarter, two failed compliance attestations in 2023, and a 14-month-old Postgres major-version upgrade nobody was brave enough to ship.

"We didn't have a platform problem. We had a platform-team problem. Twenty-two of our smartest engineers were holding the floor up instead of building floors." Julia Marchetti · VP Engineering, Helvetia Health

02The decision

Helvetia evaluated three paths in Q1 2024: stay self-hosted and double the platform team; lift-and-shift to GKE Autopilot; or move to Scalable with bring-your-own-cloud, keeping the data plane in their Zürich and Geneva regions.

Scalable won the bake-off on three things the team didn't expect to be deciding on:

  • Audit inheritance. Scalable's EU-pinned data residency, signed audit log, and DPA Art. 28 dropped four full audit controls off Helvetia's matrix on day one.
  • Two-click deploys. A real demo from a real product engineer (not a sales engineer) shipped a new service to two regions in 47 seconds. Helvetia's own pipeline took 23 minutes for the same shape.
  • Zero rewrite. BYOC meant their existing Docker images, Postgres clusters, and Kafka topics moved as-is. No Helm to rewrite, no service mesh to relearn.

03The migration

The cutover ran six weeks, end to end, with one Scalable solutions engineer embedded part-time. Helvetia moved in waves — internal tools first, then back-office, then the patient-facing services. Production traffic shifted on a Tuesday at 03:14 CET and nobody outside engineering noticed.

Before · After
LayerBeforeAfter
Computek8s on bare-metalScalable + BYOC (Zürich, Genève)
DeploysArgoCD + custom CLIscalable deploy (47s p50)
SecretsVault (self-run)Scalable Vault + BYO HSM
ObservabilityLoki + Tempo + GrafanaScalable Observability
Auditmanual evidenceinherited + signed event log
On-call3 rotations · 22 engineers1 rotation · 2 SREs

04The result

Helvetia kept two SREs on platform. The other twenty engineers moved to product squads — three new product lines shipped in the year that followed, including a long-stalled patient-portal rewrite.

The bill landed at CHF 612k for the full year — a 68% drop, even after a 22% increase in compute usage. Audits are now run by a single compliance manager pulling a signed CSV from the Scalable dashboard.

"Our last EU residency audit took forty minutes. One person, one CSV, one approval signature. The previous one took six engineer-weeks." Pavel Ritter · Staff SRE, Helvetia Health

05What's next

Helvetia is rolling out a new clinician-facing API to four EU regions in Q1 2026, on the same Scalable plane. The platform team is still two people. The audit is still forty minutes.

02The numbers

Eighteen months,
two line items.

Infra spend, CHF / month
158k → 51k
CUTOVER
Total spendCutover: Aug 2024
Engineers on platform vs. product
22 → 2
PlatformProductCutover: Aug 2024
03How it went

Six weeks,
one solutions engineer.

WEEK 0 · 22 JUL 2024

Kickoff & service inventory

Helvetia and Scalable mapped 240 services into 9 migration waves. BYOC connections to the Zürich and Genève facilities went live the same day.

WEEK 1–2 · 29 JUL — 11 AUG

Internal tools & back-office

40 non-customer-facing services moved first. Audit trail and IdP federation switched to Scalable. Two on-call rotations consolidated into one.

WEEK 3–4 · 12 — 25 AUG

Patient-facing read paths

Patient portal and clinician dashboard moved behind a blue/green flip. 60% of customer traffic on Scalable by end of week 4.

WEEK 5 · 26 AUG — 01 SEP

Postgres & the upgrade nobody shipped

Managed Postgres migration, plus the 14-month-old major-version upgrade. Zero downtime, run during a regular maintenance window.

WEEK 6 · 02 — 08 SEP

Cutover & decommission

Final 100% traffic switch at 03:14 CET, Tuesday. Self-hosted control plane archived on Friday. Twenty engineers reassigned the following Monday.

+12 MONTHS · SEP 2025

Still two SREs.

22% more compute, three new product lines, one EU residency audit completed in 40 minutes. Platform team headcount: unchanged.

"The first time the audit team called and asked if we'd actually shipped, I knew we'd made the right call. They were used to seeing six weeks of evidence collection. This time we sent them a signed CSV in forty minutes and they thought it was a mistake."
Julia MarchettiVP Engineering · Helvetia Health · Zürich

Be the next "actually shipped?" moment.

Free to start. $200 in credits. Most teams are in production within an hour — and most audits are done in forty minutes.