Steel// DEVOPS RETAINER · KUBERNETES · ON-CALL
We Eat Rusty Infrastructure for Breakfast.
Industrial-grade DevOps for teams who refuse downtime. Senior platform engineers embedded inside your stack — Kubernetes, Terraform, observability, incident response — absorbing the 2 AM pages your team shouldn't have to.
30 minutes. A senior solutions engineer — not an SDR. You leave with a written incident-readiness score and a concrete 90-day plan.
// THE 2 AM PROBLEM
Your best engineer just quit because of a pager.
Every Series B–D SaaS we audit tells the same story: a single on-call rotation is burning out the three people holding the stack together. They don't leave for the money — they leave because the postmortem lands on their desk, the vendor you hired shipped a junior, and the 2 AM page about an etcd quorum they didn't build is still ringing their phone.
A platform lead at a 90-person startup told us his team logged 410 hours of on-call in one quarter — 73% of it on infrastructure no one on the team had personally shipped. That's not on-call. That's debt collection.
SteelRats is the senior-only alternative. We don't send a junior to learn on your incident bridge. We send a platform engineer who has shipped to production at scale for eight-plus years, joins your bridge in under four minutes, and stays until the postmortem has a real action item with an owner.
// THE EMBED MODEL
Four things your SteelRats engineer owns. Nothing is outsourced. Nothing is handed to a junior.
A retainer ships one named senior platform engineer into your stack. They get repo access, prod read-only, pager seat, and a Slack channel your team actually uses. Here is the scope they pick up in week one.
-
CNCF CERTIFIED
01
Kubernetes & platform runtime
Cluster upgrades, GitOps reconciliation, multi-tenant policy, node-pool economics, secrets, ingress, cert rotation. We harden the runtime so the next CVE is a PR, not a Sev-1.
────── typical: EKS · GKE · Talos · Crossplane ──────
-
INFRASTRUCTURE-AS-CODE
02
Terraform & environment parity
We replace the snowflake dev cluster with reproducible modules, drift detection, and a merge queue that actually enforces review. No more "works on my laptop" — prod, stage, and preview converge.
────── typical: Terraform · OpenTofu · Atlantis ──────
-
OBSERVABILITY
03
Observability & SLO engineering
Metrics, traces, logs that share a schema. SLOs with real error budgets and burn-rate alerts. Dashboards your on-call can actually triage from at 2 AM — not a 200-tile Grafana wall.
────── typical: Prometheus · Grafana · Tempo · OTel ──────
-
SLA-BACKED
04
Incident response & on-call
We join your bridge within four minutes, 24/7/365. We run the incident, write the postmortem, and stay until the action items have owners and dates. Your team sleeps. We do not.
────── typical: PagerDuty · Incident.io · FireHydrant ──────
// RECEIPTS, NOT ADJECTIVES
Five years on call. 184 stacks in production. The numbers are the pitch.
- 71% avg MTTR reduction, first 90 days ▲ verified across 180+ environments
- 184 active client companies as of Nov 2024
- 97% client retention, 24-month cohort ▼ industry avg: 58%
- 14,602 production incidents resolved in 2023 ▲ zero outages > 8 minutes
- 312 AWS, GCP & CKA certifications on the team across 47 engineers
- 4min SLA: bridge join, 24/7/365 ▲ contractually backed
- #1 Best DevOps Startup — KubeCon NA 2023 CNCF, San Diego
- $14.2M Series A, Foundry Group · 2022 ▲ bootstrapped profitable Q4 '23
────── figures from internal client ledger, KubeCon jury record, and Foundry Group close announcement ──────
// WHO LETS US INTO THEIR BRIDGE
Engineers at these companies let us into their incident bridges.
Alumni-founded startups from Stripe, Ramp, and Datadog. Eleven Y Combinator W23 graduates. Press coverage in the publications platform leads actually read.
// PEER SET
- Stripe alumni-foundedx4
- Ramp alumni-foundedx3
- Datadog alumni-foundedx5
- Y Combinator W23x11
// PRESS & RECOGNITION
- The Pragmatic Engineer— issue 47, profile
- Increment Magazine— "On-call, done right"
- InfoQ— 2024 State of DevOps report
- G2 Top 20 DevOps Tools— Q1 & Q3 2024
- CNCF Member— since 2020
- O'Reilly— Incident Response Playbook, 2023
────── 184 active client companies · 97% 24-month retention · 230+ orgs use our playbook ──────
// QUESTIONS A PLATFORM LEAD ASKS BEFORE SIGNING A RETAINER
The honest answers, before the audit call.
Is this a fixed retainer or hourly billing?
Fixed monthly retainer. The number we quote in the audit is the number on the invoice. We don't bill hourly against a postmortem, and there are no surprise invoices after a Sev-1. If we exceed scope in a month, we eat it and flag it for the next planning cycle.
Do juniors ever touch our stack?
No. Every SteelRats engineer has shipped to production at scale for eight or more years. There is no offshore外包 tier and no "shadow" bench learning on your incident bridge. Your named engineer is a senior, and we keep a second senior on standby for PTO coverage so continuity never drops to a junior.
What's your actual incident SLA?
A SteelRats engineer joins your incident bridge within four minutes of paging, 24/7/365. That is contractually backed — not a marketing number. We do not promise sub-one-minute response because we won't write a check we can't clear.
Are you replacing our in-house platform team?
No. We augment. Most clients keep one or two in-house platform engineers; we act as the senior force-multiply and the on-call safety net. At the end of every quarter, the system is yours — runbooks, Terraform modules, dashboards, all of it. We make ourselves unnecessary, then we stay anyway because the retainer is worth it.
What does the free platform audit actually look like?
Thirty minutes with a senior solutions engineer — not an SDR. We come having read your stack page, your status page history, and your last two public postmortems. You leave with a written incident-readiness score (six categories), a 90-day plan with named work, and a candid yes/no on whether we're the right fit.
What's your security posture today?
SOC 2 Type I as of Q2 2024, with Type II in progress. Type I covers the controls; we won't call ourselves Type II until the audit window closes and the report lands. We are happy to send the SOC 2 package under NDA before the audit call.
Still on the fence? The audit call is the easiest way to find out.
Book Your Free Platform Audit →