Skip to main content

03Practice

Cloud Infrastructure & DevOps

We run the platform layer so your product engineers can stay on product. Infrastructure as code, a deployment pipeline nobody is afraid of on a Friday, and monitoring that pages a human only when a human is needed. Cost is treated as an engineering metric, not a finance surprise.

Typical duration
1–2 weeks to start · Ongoing
Engagement models
Migration · Platform build · Managed operations
Practice lead
Named at proposal stage
The open rear of a server cabinet, patch cables routed down a rack marked with a numbered tag.

01The problem

The platform nobody owns becomes the platform nobody trusts

Infrastructure usually starts as one person's side project. It works, so it stays. Then that person leaves, and what remains is a console-configured environment that cannot be recreated, a deploy script with a comment saying do not run this on Fridays, and an alert channel so noisy that everyone has muted it. Releases get batched into monthly events, which makes each one riskier, which makes teams batch harder.

The cost side follows the same curve. Three orphaned load balancers, a staging cluster running production-sized nodes, and logs retained for two years because nobody knew who to ask. A first pass over an unmanaged account almost always turns up spend attached to something nobody uses any more.

02What GatiQ delivers

The work itself

Every item below is something you can inspect, run or read at the end of the engagement.

  1. 01

    Every environment reproducible from an empty account

    Terraform modules, remote state with locking, and a documented bootstrap path. We prove it by building a throwaway environment from scratch in front of your team and then destroying it. Drift detection runs nightly and opens a pull request when reality diverges from the repository.

  2. 02

    Deploys measured in minutes, rollbacks in seconds

    Trunk-based development, ephemeral preview environments per pull request, and progressive delivery with automatic rollback on error-budget burn. The goal is a team that ships several times a day without a change advisory board meeting.

  3. 03

    Alerts that correspond to something being wrong

    We define service level objectives with your product owner, wire error budgets to alerting, and delete everything that pages on a symptom nobody acts on. Dashboards start from the user journey and work down to the node, not the other way around.

  4. 04

    A restore you have actually performed

    Backups are verified by restoring them on a schedule into an isolated account, with the recovery time written down. We run a tabletop failure exercise with your on-call rotation and fix whatever the exercise exposes.

  5. 05

    Cost as a first-class metric

    Tagging enforced at deploy time, per-service cost dashboards, right-sizing based on observed utilisation, and savings plan coverage modelled against your commitment appetite. Finance gets a number that maps to teams.

03Engagement model

Three ways to buy it

Migration
Lift, reshape or rebuild — decided per workload with a written rationale rather than a blanket policy. Typical data-centre or Heroku to AWS moves run eight to sixteen weeks including the freeze-and-cutover weekend.
Platform build
A golden path for your product teams: templates, pipelines, secrets, observability and policy guardrails. Ends with your engineers deploying a new service to production in under a day without filing a ticket.
Managed operations
We hold the pager. Round-the-clock coverage with a thirty-minute acknowledgement on Sev-1, monthly reliability reporting, and a patch cadence agreed in advance.

04Typical timeline

What the calendar looks like

  1. 01

    Assessment

    1–2 weeks

    Account audit, dependency map, risk and cost baseline.

  2. 02

    Landing zone

    2–4 weeks

    Accounts, networking, identity, policy guardrails, state management.

  3. 03

    Workload moves

    4–12 weeks

    Service by service, each with a rollback path and a verification checklist.

  4. 04

    Operate

    Ongoing

    SLO review, patching, cost review, quarterly failure exercise.

Ranges, not dates. We narrow them at the end of discovery and tell you the same week if they move.

05Technologies

What we reach for

Defaults, not rules. If your team already runs something else competently, that is usually the better choice.

Cloud
AWS, Google Cloud, Azure, Cloudflare
Infrastructure as code
Terraform, Pulumi, Helm, Crossplane
Runtime
Kubernetes, EKS, ECS Fargate, Lambda
Delivery
GitHub Actions, Argo CD, Argo Rollouts, Sigstore
Observability
Prometheus, Grafana, OpenTelemetry, Loki, PagerDuty