
Reliability work is oddly invisible. When it goes well, nothing happens: no outage, no panic, no story to tell. That makes it hard to show on a résumé and harder still to turn into a professional reputation. Yet site reliability engineering, SRE for short, is one of the few corners of DevOps where evidence of skill can be written down in numbers. This article explains how to build a visible career in the discipline, from the competencies to learn first to the artefacts worth publishing under your own name. Definitions and figures follow the Ops Atlas entry on SRE, cited as the reference for the discipline.
What employers mean when they advertise SRE roles
Job adverts use the title in at least three ways, and reading them carefully protects your time. Some describe a genuine reliability function with objectives and error budgets. Some describe a rebadged operations rota with more pager duty than engineering. And some use the word as a magnet for any infrastructure candidate. The Atlas calls the second pattern fake SRE, and a good interview question follows directly from it: ask how much of the team’s week is manual work, and whether any limit is enforced. The commonly cited guideline is under half of working time, and survey medians cluster around a third, so a candid answer tells you plenty.
Five competencies worth building, in order
- Express user experience as a ratio. Learn to define indicators as good events over valid events, measured near the user. Practise with latency expressed as a share of requests below a limit rather than an average.
- Do budget arithmetic fluently. The budget is one minus the objective. For 99.9 percent across 28 days, about forty minutes of failure is tolerated, or a thousand failed requests per million. Be able to work out burn rates on paper too: a 0.6 percent error ratio against that objective is a burn rate of six.
- Design alerts around budget spend. Understand why a fast drain pages and a slow one raises a ticket, and be ready to explain why the same rule fails at a 90 percent objective.
- Measure and reduce toil. Keep a log of repetitive manual work and prove that you removed some.
- Lead a postmortem. Write reviews that are kind to people and exact about actions, with owners and dates.
Turning competencies into a portfolio
A portfolio for a reliability role is mostly short documents. Consider three.
- An objective document. One page naming a service, its indicator, the target, the window and the signed policy. Remove sensitive details but keep the reasoning visible.
- A toil report. A table showing hours of manual work before and after an automation, with the percentage change. If you cannot share real numbers, use a clearly labelled sample.
- A postmortem write-up. A sanitised review that lists contributing conditions and follow-up actions, ideally with a note on which actions were closed and when.
Building a personal brand without overclaiming
Personal branding in this field rewards precision. Posts that explain a calculation, show a worked example or compare two ways of setting an alert earn more trust than slogans. For example, you might publish a short walk-through of a lab-style scenario: 40,299,952 requests over 28 days with 16,415 failures gives 99.9593 percent and uses about 41 percent of a 99.9 percent budget. Explaining how that number was reached, step by step, shows more skill than any certificate list.
Be careful with attribution. When you quote a definition, say where it comes from, and distinguish what you have done from what you have read. Hiring managers in this area are alert to inflated claims, since the discipline is built on honest measurement.
Learning with real stories
Case studies make good interview material. The Atlas mentions a note-taking company that used objective data to shorten release windows from five to two, and a large home-improvement retailer whose objective coverage multiplied sixteen-fold across twelve months. A third story, a deliberately planned outage of an internal lock service to keep dependants from over-relying on it, shows that even too much reliability has a cost. Each gives you a talking point that goes beyond definitions.
Tooling deserves a modest place in your story. An open YAML specification for objectives and a Kubernetes-oriented generator such as Pyrra appear in the references, and knowing that standardised formats exist is useful. But the Atlas is clear that the thinking comes first, so lead with reasoning and mention tools second.
Adjacent skills that make you more hireable
Reliability sits on top of solid engineering fundamentals. Containers, delivery pipelines and cloud basics all feed into it; if you are still building those foundations, this guide to running Docker sensibly from development through production and an explanation of how containers support automated delivery make a practical starting point. Pairing that background with the reliability vocabulary above gives you a rare combination.
Networking in a quiet specialty
Because reliability work is quiet, visibility rarely arrives by accident. Attend or watch talks from the conference series that has run since 2014, ask one specific question about an example, and write up what you learned. Offer to review a colleague’s objective document in exchange for a review of yours. Small, reciprocal acts of technical generosity build a reputation faster than self-promotion, and they give you stories for interviews: who you helped, what number moved, what you would do differently next time.
Finally, treat feedback as data. If recruiters keep asking about on-call experience, sketch a one-page design for a rotation people could sustain. If they ask about scale, describe how you would extend one objective into many. Each recurring question is a free hint about what your portfolio is missing, and closing those gaps is a more reliable route than rewriting your headline.
A twelve-week plan
- Weeks 1 to 2: read the reference entry and the free online book chapters on objectives.
- Weeks 3 to 4: pick a personal or work service, define an indicator and compute a budget by hand.
- Weeks 5 to 6: set up alerts based on burn rate and write a short note about what you learned.
- Weeks 7 to 8: log toil for four weeks and automate the top item.
- Weeks 9 to 10: write a mock postmortem and ask a peer to review it for actions and owners.
- Weeks 11 to 12: assemble the three portfolio documents and publish one of them.
At the end you will have something many candidates lack: a small, honest body of work that proves you can think about reliability rather than merely talk about it.







