Senior engineers stay accountable for every change. AI drafts the IaC, combs the logs, and writes the first runbook — under a written human review gate.
Perun Engineering is an Australian consulting practice, headquartered on the Gold Coast with senior engineers across Canada and the EU. Engagements follow the sun, led end-to-end by a principal — no offshore handoff, no junior rotation, no reselling of someone else's reference architecture.
We are an AWS Partner Network member. Our Australian team holds every active AWS certification (AWS Golden Jacket), and includes AWS Community Builders and AWS User Group leaders. That's not a decoration — it means you get people who contribute to the ecosystem they deploy into.
Our Australian principals travel on-site when the engagement calls for it — Sydney, Melbourne, Brisbane, Perth, Adelaide, Canberra. Remote-first by default, on-site when the work, the compliance regime or the room demands it.
Every principal has shipped inside PCI DSS, SOC 2, ISO 27001 and HIPAA environments — and also inside early-stage startups where velocity over complexity is the only way to survive. We know when to add the control, and when to remove it.
AI is not a service we sell. It is how we deliver the services we already sell. Every engagement runs through the same written review gate — AI drafts the IaC, the runbook, the log triage; a named senior engineer signs the diff; the gate is the artefact you keep. We will not bill you for an AI tier, and we will not hand you autonomous agents. Both are the wrong unit of trust.
A room of principals, billing hourly, typing production configs by hand. It works — at a price most teams can't sustain past the first quarter.
Every change is a written diff signed by a named senior engineer. AI drafts the IaC, combs the logs, and writes the first runbook — the human owns the call. The review gate is the durable artefact: the bit you keep, and the bit a regulator can read.
Autonomous agents let loose on your cloud. Fast on paper — until the first 3am incident nobody can explain and the audit trail is a chat transcript.
All three practices ship the same way: senior engineers sign every diff, AI drafts the rest, the review gate is written down.
Multi-account landing zones with guardrails, cost visibility, and compliance scaffolding. Re-platforming projects where getting the account structure right matters more than clever tooling.
Opinionated internal platforms on EKS: GitOps with Argo, progressive delivery, secrets done properly, autoscaling that doesn't flap. We build the platform team as we build the platform.
Evaluation harnesses before model work. Retrieval pipelines that are boring in the places that matter. Inference at a predictable p95. We help teams move from demo to durable.
Straight answers about what Perun does, where it is based, and how it ships. Choosing a partner? Read our guide on how to choose an AWS consulting partner in Australia.
CIDR redesign and fresh-VPC cutover run under live payments traffic, then Transit Gateway retired for a single Cloud WAN hub. RDS, EKS, OpenSearch and ElastiCache replatformed with zero downtime; multi-region, multi-cloud and on-prem traffic now share one fabric.
Cutover from Ingress NGINX to a supported controller, run against a fleet of white-label tenant domains under live payments traffic. The migration was driven by a hard upstream end-of-life date and could not take a maintenance window — both controllers ran side-by-side while traffic shifted progressively, with no customer-visible disruption.
Modernised an over-stretched multi-cluster Prometheus stack into a highly available setup, with Thanos handling long-term metrics storage. The platform scales horizontally, ships with SSO and access control at the Grafana layer, and onboarding a new cluster is a single remote-write target.
Built a PCI-aligned path for engineers to reach internal resources across the entire AWS organisation. Bastion images, identity configuration and audit are centralised and distributed org-wide; every shell is tied to a real corporate identity through SSO, with privileged bootstrap rights surrendered as soon as the host is up.
Cutover from a legacy Bitbucket and Jenkins SDLC onto GitHub and GitHub Actions for an IoT engineering team, then layered the platform with secured MCP servers on Amazon Bedrock AgentCore, standardised Claude Code and OpenCode skills and tools, and a management view of token spend. Engineers get faster AI-assisted troubleshooting against real infrastructure state; leadership gets per-team visibility into AI usage and cost.
The 2-hour cutover window with zero downtime. We had live payments traffic flowing throughout — we'd planned for a full maintenance weekend.
Scaling the monitoring estate used to mean weeks of config work. Now adding a cluster is a single config change. It finally keeps up with how fast we're moving.
Our engineers now have safe, auditable access to live infrastructure state. Working against production signals instead of documentation changes what's possible.