Site Reliability Engineer

Anyfin; Stockholms kommun; On-site; SEK 700,000 - 900,000; Full time

Job summary

Anyfin is seeking an experienced SRE to own reliability across our Kubernetes/GCP-based platform. You will implement monitoring, alerting and logging to detect issues before customers are affected.

You’ll drive incident response, blameless postmortems and cost-conscious scaling, while shaping CI/CD and infrastructure as code practices. You’ll collaborate with backend and data teams in a fintech environment, ensuring resilience for a regulated lender.

Responsibilities

• Own reliability across the platform and services running on Kubernetes and GCP.
• Implement monitoring, alerts, and logging to surface issues before customers notice.
• Shape incident response, on-call rotation, and postmortems to drive real changes.
• Contribute to CI/CD and infrastructure as code to enable safe, fast shipping.
• Ensure regulatory and bank-related resilience requirements are met, including cost awareness.

Skills

SRE experience

Kubernetes

Infrastructure as code

Observability

On-call

Go

Python

TypeScript

Tools

Kubernetes

Terraform

GCP

Job description

When Anyfin goes down, someone’s loan payment doesn’t go through. Make sure it doesn’t.

You open your laptop on Monday and find out the primary database failed at 3am on Sunday. Services reconnected in seconds and no customer noticed a thing. We’d practised exactly that failover the week before.

If that’s your idea of a good weekend, we should talk.

The foundation is built. Now it needs to scale.

Anyfin helps people refinance expensive consumer credit on better terms. Since 2017 we’ve helped people across Sweden, Finland and Germany take back control of their finances. We’re profitable, we have a banking licence that lets us operate across the EU, and we’re getting ready to launch new products and enter new markets. Our ambition is to be a pan-European company by 2030.

Every one of those steps puts more weight on our platform: more traffic, more integrations, more markets and more regulators who expect us to prove that we’re resilient, not just say it.

The team is small and intentionally so. We’d rather grow what each engineer can do than how many engineers we have.

The actual job

You’ll own reliability across the platform. That covers how our services run on Kubernetes and GCP, how we deploy, how we scale and how we recover when something breaks. You’ll decide what “reliable enough” means for each part of the system and make it measurable.

You’ll make problems visible before customers feel them. That means monitoring, alerting and logging that tell us what’s wrong and why, not dashboards nobody opens. When an alert fires, it should matter.

You’ll make incidents boring. You’ll shape how we respond, who gets paged, how we communicate and how we learn. That means blameless reviews that lead to real changes, not just documents.

You’ll build the paved road. Engineers should be able to ship safely and quickly without thinking about infrastructure every time. You’ll work on CI/CD, infrastructure as code and the tooling that makes the right way the easy way.

You’ll help us meet the bar a bank is held to. As a regulated lender, operational resilience isn’t optional. You’ll help us turn requirements like DORA into systems and practices that actually work, not just policies we can point to.

You’ll keep an eye on cost. Scaling into new markets shouldn’t mean scaling our cloud bill at the same rate.

Reliability in an AI-native company

AI isn’t a side tool here. It’s built into how we work across engineering, analytics, operations and product, and our tech team has built an AI-connected data and code environment on frontier models.

That shows up in reliability work too. AI can help read logs, trace incidents and draft runbooks faster than any of us. It also means more systems in production act on data without questioning it, and more automation can change things on its own.

Part of this role is working out what reliability means in that world: where AI makes us faster, where it needs guardrails, and how we make sure we’re never the last to know when something goes wrong.

The people

You’ll work closely with backend engineers, data engineers, product teams and our engineering managers, who care a lot about getting things right. Sometimes a little too much. Propose a new alert and someone will ask how many times it would have fired last month. We prefer it that way.

They’re also generous with what they know, quick to laugh and happy to change their minds when you make a better case. As long as you can back it up.

This sounds like you if…

You have solid experience in SRE, platform or infrastructure engineering, ideally in a product company where reliability directly affects customers. You’ve run production workloads on Kubernetes in a cloud environment and are comfortable with infrastructure as code.

You write real code, not just YAML. Our main stack is TypeScript, Node.js, Python and Go, running on Kubernetes and GCP. Experience with GCP and Terraform is a plus, but you don’t need to know every technology in our stack. You’re comfortable reading, debugging and improving services, and the judgment you bring matters more than ticking every box.

You’ve built observability that people actually use, and you’ve been on call for systems that mattered. You know the difference between an alert and noise.

You’re pragmatic. You care about reliability, but you know 100% uptime isn’t the goal and that over-engineering has a cost too. You can explain that trade-off to engineers and non-engineers alike.

Experience from fintech, banking or another regulated environment is a plus, but not a requirement.

More than anything, you can tell us about a system you made more reliable. Not just what you changed, but what it made possible for the team and the business.

A few practical notes

This role is based at our office on Drottninggatan 92 in Stockholm, and we work from the office. We think great teams and culture are built when people meet in person, especially at this stage.

EU work authorisation is required.

If you’d rather prevent the next incident than write the postmortem for it, we’d love to hear from you.