Site reliability engineering for UK SaaS and fintech — SLOs, error budgets, incident response, runbooks, on-call. We stop the 3am pages by building the discipline and the platform that prevent them.
Reliability isn’t a tool you buy. It’s a discipline you build: SLOs that match the business, error budgets that decide when to ship, on-call that doesn’t burn your seniors out.
The in-house attempt was two engineers part-time, no observability, no audit log, and a Confluence page with the runbook on it. When the acquirer’s CTO walked the codebase pre-LOI, the verdict was blunt: “rip and replace, or a 25% diligence haircut.”
The product wasn’t the problem. The reliability discipline was. SLOs didn’t exist. Backups ran but restores were never tested. One engineer knew the system, so the bus factor was one. Nothing about it was defendable on a diligence call.
We rebuilt over 10 to 14 weeks. Senior engineers paired daily, ADRs landed the same day, the evidence pipeline was wired on day one. The acquirer’s CTO walked the same surface, signed full-team retention, and the deal closed at ask. This page is for the CTO who decided reliability is the product floor.
Reliability is bought as a tool by everyone who hasn’t felt a 3am page. It’s built as a discipline by the people who have. Here’s what built looks like.
The engineers who sign your scope are the engineers who ship and maintain. Names on the SOW are names on the commits. No staff aug, no day-rate, no rebuild upsell.
Every decision in an ADR. Every action in an audit log. SLOs and error budgets documented. Acquirer CTO and regulator both defendable on the call.
Every workflow rerunnable. State recoverable. Disaster recovery rehearsed quarterly with a documented RTO and RPO, not a backup nobody’s ever restored from.
An £8K five-day audit, then a fixed-price sprint quoted at audit end. No scope creep. 30-day walk-away both ways. IP assigns on commit. No deposit forfeiture.
The discipline that turns a fragile platform into one your on-call can sleep through. Audit-led, evidence-first, observable from day one.
Service level objectives that match what customers feel, not a vanity 99.99%. Error budgets that decide when to ship and when to freeze.
Trace IDs end to end, PagerDuty wired to the right humans, a runbook per alert. On-call reads the trace and knows in 60 seconds.
A DR drill on the calendar, RTO and RPO tested, restore under four hours. We break things in a controlled window so they don’t break at 3am.
A rotation that’s humane, alerts that are actionable, blameless postmortems. The muscle stays with your team after we’ve gone.
Observability as the source of truth. Infrastructure as code. The hiring pool that means your next engineer is productive on day one.
Each one is the difference between a platform your on-call sleeps through and one that pages them awake.
Production incidents since cutover
SRE engagements since 2019
“Two senior engineers, five days, and a reliability plan you can take to your board. That’s where every engagement starts.”
A workflow died silently and the customer found out first. Per-workflow observability, trace IDs end to end, PagerDuty and Slack wired to the right humans.
Who did what, when? No record, and the SOC 2 auditor flagged it. Every change logged, tenant-admin viewable, exportable. Regulator request defendable inside the deadline.
The cloud bill spiked and nobody knew why for three weeks. Cost tags per workload, budget alerts to Slack. The bill stabilises within 30 days and you can name what every pound buys.
The backup runs, the restore has never been tested. A DR drill on the calendar, RTO and RPO documented and tested, restore in under four hours, proven on a real run.
One engineer knows the whole system, bus factor one. Two senior engineers paired daily, decisions in an ADR the same day. Bus factor moves from one to four.
On-call gets 40 pages a night and ignores all of them. A noise budget and actionable alerts. Every alert maps to a runbook and a human action. Fewer pages, all worth waking up for.
Production incident at 3am and on-call has no idea what happened. Trace IDs end to end, structured logs. On-call reads the trace and knows the root cause in 60 seconds.
We rebuilt a fragile, undocumented platform as one an acquirer’s CTO could defend: SLOs and error budgets, full observability, a tested DR plan, an evidence pipeline wired from day one.
Two phases, risk bounded at week one. You decide whether to continue after a fixed-price audit, not after a 12-week SOW.
Two senior engineers read your estate and brief, then hand you the receipts to decide.
Eight to fourteen weeks of fixed-scope shipping. Audit-led, evidence-first, observable.
One day a week of senior reliability work for three to six months while your team gets unblocked.
Pain-first, soft-second.
An £8K five-day audit, then a fixed-price sprint typically £45K to £110K depending on the estate. The sprint is scoped at the audit, so the number you sign is the number you pay. No day-rate, no scope creep, no surprise invoice in week ten.
No. We work in place where possible. If a migration genuinely helps, it gets scoped at the audit and you decide. Old and new run in parallel until you sign off. We don’t rip and replace for the sake of a rebuild invoice.
An ADR pack, an audit log, SLOs and error budgets documented, an evidence pipeline, and a SOC 2 stance. The acquirer’s CTO walks the surface and signs full-team retention. We’ve done exactly this three times across 2024 and 2025.
Every engagement has two senior engineers paired, not one. Every decision goes into an ADR the same day. Bus factor is four, not one. Two handovers in seven years, both inside 48 hours. The names on your scope are the names on your commits.
Yes. AWS is the default. We’ve shipped reliability work on GCP, Azure, Cloudflare, DigitalOcean, Vercel, and Fly.io. The multi-cloud lock-in path gets documented so a future board demand for a different provider isn’t a rewrite.
A Vanta or Drata evidence pipeline from day one. SOC 2 Type 1 achievable in eight to twelve weeks. A signed DPA plus SCCs for cross-border. A DSAR runbook and an erasure tool honoured at the schema. Procurement clears in days.
Yes. A good chunk of our reliability work is two seniors pairing with your team, not replacing it. Your engineer learns the on-call discipline, the SLO thinking, and the runbook habit while we ship. When we leave, your team owns it.
Walk away whenever. 30-day notice both ways, IP assigns on every commit, no lock-in, no deposit forfeiture. The retainer is optional and month to month. You keep everything we ship the day you stop paying.
A look at the kind of site reliability engineering surface we hand over — real screens, real data, documented and yours from day one.
Tell us your current estate, your deadline, and the outcome you need. Mohit replies inside 24 hours: a clear yes, a clear no, or the one question that decides it, plus your next audit slot.
Yes, no, or the deciding question. Straight to your inbox, with an audit slot.
30-page brief, six ADRs, a ranked risk matrix, and a fixed-price quote for the sprint.
SLOs, observability, runbooks, DR tested. A platform your on-call can sleep through.