~/services
// 01 · services

Three services. One team that operates what it builds.

No hand-offs to a support desk that has never seen your system. The engineers who design and build your infrastructure are the ones on call for it. Here's what that looks like in practice.

~/operack — services
$ operack services --list
  devops     pipelines, iac, kubernetes, on-call you can sleep through
  infra      architecture, migration, cost & security reviews
  build      backends, apis, data pipelines, product mvps
$ engagement model: senior-only · fixed-scope or retainer
01 / devops

DevOps & Platform

We make deploys boring — and keep them that way.

Ship on demand, roll back in seconds, and get paged only when it actually matters. We build the pipeline, the platform and the alerting as one system — not three tools bolted together.

deploy.log
build ................. ok 42s
tests ................. ok 318 passed
canary 5% ............ healthy
promote 100% ......... ok
deploy #1284 → prod ● live
// ci-cd

CI/CD & release automation

Pipelines that build, test, canary and promote — with fast, safe rollbacks and no manual gates you'll forget.

// iac

Infrastructure as code

Terraform or Pulumi with GitOps. Every change reviewed, versioned and reproducible — no click-ops drift.

// k8s

Kubernetes & containers

Right-sized clusters, sane defaults, autoscaling and secrets done properly. Or nothing at all if you don't need it.

// observability

Observability & alerting

Metrics, logs and traces wired to alerts that mean something. SLOs, dashboards and runbooks included.

// on-call

On-call & incident response

Escalation, runbooks and blameless post-mortems — so 3am pages get rarer, not louder.

// hardening

Supply-chain & hardening

Signed builds, scanned images, least-privilege access and secrets rotation baked into the pipeline.

02 / infrastructure

Cloud Infrastructure

Only what your workload needs — no buzzword bill.

Architecture that fits the load, migrations that don't lose sleep, and a cloud bill you can actually explain. We design for reliability and cost together — because in production they're the same conversation.

cost-review.txt
before ............. $18,400 / mo
right-sizing ....... −$6,200
commitments ........ −$3,100
idle cleanup ....... −$1,450
after .............. $7,650 / mo −58%
// architecture

Cloud architecture

Reference designs for compute, data and networking — documented, diagrammed and sized to your traffic.

// migration

Migrations & modernisation

Lift-and-shift or re-platform, with cutover plans, dual-running and a tested rollback at every step.

// multi-cloud

AWS · GCP · Azure · bare metal

We're not tied to one vendor. We pick where each workload runs cheapest and most reliably — including your own hardware.

// cost

Cost & right-sizing

Find the waste, commit the baseline, and set up guardrails so spend doesn't creep back.

// security

Networking & security

Segmentation, private connectivity, IAM and encryption — reviewed against a real threat model, not a checklist.

// reliability

Reliability reviews

Failure-mode analysis, backups you've actually restored, and DR you've actually tested.

03 / development

Software Development

Built from day one to be deployed and operated.

Backends, APIs, data pipelines and full products — written by people who will also put them in production. That means logging, metrics and failure handling aren't an afterthought bolted on before launch.

GET /api/v1/health
HTTP/1.1 200 OK
status ......... "ok"
db ............. "connected" 2ms
queue .......... "drained"
build .......... "v2.4.1"
// backend

Backend & API engineering

Typed, tested services with clean contracts — REST, GraphQL or events. Documented and versioned.

// data

Data & event pipelines

Ingestion, transformation and delivery you can trust — with backpressure, replay and observability built in.

// tools

Internal tools & dashboards

The unglamorous admin panels and ops tooling that make a team faster. Shipped, not prototyped.

// mvp

Product & MVP builds

Zero to production for a new idea — with the infra, CI/CD and monitoring already in place at launch.

// integrations

Integrations

Third-party APIs, webhooks and messaging wired reliably — with retries, idempotency and dead-letter handling.

// eo

Geospatial & earth-obs

Satellite and sensor data — Sentinel, Copernicus — turned into pipelines and products. See BOABE.

04

How an engagement runs

Same four steps whether it's a two-week audit or a six-month build.

step 01

Scope

A short, honest conversation. We map the problem, the constraints and whether we're even the right team. No charge.

step 02

Plan

A written proposal — scope, milestones, price and who does what. Fixed-scope or a monthly retainer, your call.

step 03

Build & operate

We ship in small increments you can see, with monitoring and runbooks from the first deploy — not the last.

step 04

Hand over

Documentation, access and a clean handover — or we stay on to run it. You're never locked in to keep us.

05

Tools we reach for

Boring, proven, and only when the workload calls for it.

cloud
AWS · GCP · Azure · Hetzner · bare metal
iac & ci/cd
Terraform · Pulumi · GitHub Actions · GitLab CI · ArgoCD
runtime
Kubernetes · Docker · Nomad · serverless where it fits
observability
Prometheus · Grafana · OpenTelemetry · Loki · Sentry
languages
Go · TypeScript · Python · Rust chosen per job, not per fashion
data
PostgreSQL · ClickHouse · Redis · Kafka · object storage

Not sure which one you need?

Most projects touch all three. Tell us the problem and we'll tell you where to start — and whether it's even worth doing.