SRE / Production Engineer
An SRE / Production Engineer owns service reliability: SLOs and error budgets, incident response, on-call, and the resilience patterns (retries, circuit breakers, graceful degradation) that keep other teams' services up. This isn't sysadmin work with a rebrand — the hiring bar consistently asks for real coding and automation experience (Python shows up far more than Go, despite the two often getting equal billing), and comp bands track software engineering ladders, not IT/ops ladders. Most real postings describe the job in plainer language than the "SRE Book" vocabulary suggests, too — you'll see "own availability" and "define reliability targets" far more often than the literal terms "SLO" or "error budget," even though it's the same underlying practice. It's also not "systems work in the abstract" — this role specifically rewards people who enjoy incident composure under real time pressure. A growing slice of these postings, at places like OpenAI, Coinbase, and Cerebras, frame the job specifically around keeping AI training/inference infrastructure up rather than classic web services — worth knowing if you're eyeing an AI-native company, since the on-call texture is the same but the failure modes are different. If incident response itself doesn't appeal to you, platform engineering is probably the better fit even though the tooling looks nearly identical.
What matters most for this role
Defining trait: 24/7 follow-the-sun on-call rotations, live-outage escalation ownership; misconceptions section states the role 'rewards people who specifically enjoy incident composure.'
Owning escalation during live outages and driving MTTD/MTTR down under real time pressure is core to the role.
Central success metric is service availability against SLOs/error-budget adherence — a named, measured production outcome even when upstream code is owned by other teams.
Senior SREs influence architecture and standards broadly, designing resilience patterns across services owned by other teams — strong but reliability-focused rather than platform-building.
2025 SRE Report shows a rising 30% operational-time share; daytime work is regularly interrupted by alerts/incidents even outside formal on-call windows.
A day in this role
Google's own doctrine caps operational/reactive work at half your time, reserving the rest for project work — building monitoring and alerting tooling, defining SLOs, and driving resilience patterns across services other teams own. In practice that split slips: the 2025 SRE Report found a median of 30% operational time, up from 25% the year before, so ask specifically about the on-call load and project/ops ratio in any interview rather than assuming either extreme. Day to day you're not just watching dashboards from outside — you're building and operating the platform other teams' services actually run on: Kubernetes clusters, multi-cloud deployment infrastructure, and CI/CD pipelines wired through Terraform, Helm, and ArgoCD so releases stay safe and repeatable at scale. You instrument the monitoring and alerting yourself and use it to push MTTD/MTTR down and alert noise out, not just consume someone else's dashboard. You'll carry on-call rotations that can run 24/7 follow-the-sun across regions, own escalation during live outages, and turn what postmortems surface into automation or architecture fixes rather than just filing a report. At senior and staff levels, expect a named domain of ownership — networking/service mesh, the database/storage layer, security posture, cost/capacity — rather than generic on-call duty across everything; this is an engineering discipline, not a ticket queue.
Comp structure
Typical: $193K
Base-heavy with equity, tracking standard software engineering ladders at the same company rather than an IT/ops pay scale. On-call is generally not paid as separate variable comp at most big tech companies — it's folded into base/equity/bonus — though some companies do pay explicit stipends, and this varies enough that it's worth asking directly rather than assuming either way. levels.fyi-reported total comp for the title runs from roughly $197K (Google L3) up to $768K (Google L7), with an overall reported median around $203,600 across companies. Current live postings show that same spread by company stage and mission more than by title: mid-level roles at large regulated employers can post as low as $96K-$163K (Mastercard) or $117K-$177K on a government-cleared track (Salesforce), while senior/staff roles at fast-growing, venture-backed companies commonly list $200K-$300K+ (Harvey, EarnIn, Crusoe) — where you land depends heavily on who's hiring, not just the level in the title.
▸ Data notes▾ Data notes
▸ Full compensation breakdown by level and company tier▾ Full compensation breakdown by level and company tier
Compensation by Company Tier
Total compensation (base + bonus + annualized equity) across five company tiers, at each career level. The same role pays very differently depending on where you take it.
sre-production-engineer · total comp (base + bonus + annualized equity) · P25–P75 band, P50 median
Equity Reality Check
The guaranteed money (base + bonus) against the equity upside. Startup equity is illiquid — the equity figure is annualized paper value at vest, not cash in hand.
Examples of real job postings
snapshot from 2026-07-12Real postings from the research corpus behind this archetype. Click one to read the actual listing.
How to test this cheaply
Ask an SRE at a company you're interested in if you can shadow one on-call shift (even passively, watching how they triage an alert) to see how it actually feels under a live page.
Separately, try running a small "game day" for yourself — deliberately break something in a side project or staging environment and time how you handle the diagnosis and fix under self-imposed pressure — it's a decent proxy for whether incident composure is something you enjoy or just tolerate.
Do this role, or hire for it? Rate how much each trait actually matters. Role-holder and hiring-manager ratings are kept separate, and no single rating changes the model; ratings are aggregated with anti-gaming thresholds before they factor in.