Who we are

About the role

We are looking for an SRE / DevOps engineer who can help us make delivery faster, safer, and less dependent on manual operations work.
For the first three to six months, the focus will be reducing friction around deploying our new system, improving the reliability of our current AWS-based platform, and introducing agentic engineering practices that help engineers diagnose, fix, and safely ship changes without getting blocked on operations.
This is a hands-on production role. You will work with real infrastructure, real deployment pipelines, real incident patterns, and real engineering teams. You should be comfortable operating production-grade systems with the right checks and balances: VPN access, least-privilege permissions, runbooks, repo instructions, service maps, deployment procedures, audit trails, and rollback paths.
You will also help us use AI agents in a practical way: not as magic automation, but as supervised engineering assistants that can inspect code, read operational context, suggest fixes, generate runbooks, improve CI/CD, and help engineers debug production issues safely.

Your tasks

In the first three to six months, you will:

• Improve the deployment experience for our new system.
• Reduce operational bottlenecks that slow down engineering and feature delivery.
• Strengthen our AWS production setup, currently based on ECS and containers.
• Improve our GitHub Actions CI/CD workflows.
• Work with Terraform / OpenTofu to make infrastructure safer, clearer, and easier to change.
• Improve production debugging across AWS, containers, networking, Linux, and application-level issues.
• Improve our observability across the three pillars: metrics, logs, and traces.
• Create or improve runbooks, repo instructions, service maps, deployment guides, and operational documentation.
• Introduce agentic engineering workflows that help engineers diagnose issues, propose fixes, and validate changes before they reach production.
• Design safe guardrails for agent-assisted work: permissions, approval gates, auditability, sandboxing, rollback procedures, and human review.

Current and future platform context

Today, we run on AWS and ECS. Strong AWS experience is required.

You should be comfortable with:

• AWS production environments
• ECS and containerized services
• GitHub Actions
• Terraform and/or OpenTofu
• CI/CD pipelines
• Linux and networking
• Production debugging
• Secrets, VPNs, access control, and operational security
• Observability across metrics, logs, and traces

We are also considering a move toward Kubernetes, so Kubernetes experience is highly valuable. You do not need to migrate us on day one, but you should be able to reason clearly about when Kubernetes helps, when it adds complexity, and what a safe migration path would look like.

Your profile

We are looking for someone who has moved beyond simply “using ChatGPT.”
You should be able to use AI agents as part of a serious engineering workflow. That means you can:

• Break a large operational problem into smaller agent-assisted tasks.
• Give agents the right repo context, runbooks, constraints, and success criteria.
• Use agents to inspect code, infrastructure, CI/CD, logs, and service behavior.
• Verify agent output instead of trusting it blindly.
• Turn useful agent output into production-grade changes.
• Design workflows where agents can suggest fixes, but humans approve risky actions.
• Know when not to use an agent.

Examples of useful agentic workflows include:

• An agent helps diagnose a failed deployment by inspecting CI logs, ECS task events, recent commits, and application logs.
• An agent proposes a Terraform or GitHub Actions change, but only as a pull request with tests and rollback notes.
• An agent generates or updates a runbook after an incident, using evidence from logs, traces, metrics, and deploy history.
• An agent creates a first-pass service map from repositories, infrastructure definitions, and runtime configuration.
• An agent helps reproduce a production issue in a safe sandbox before any production change is made.

Required experience

• Strong hands-on experience in SRE, DevOps, platform engineering, infrastructure engineering, or production operations.
• Production AWS experience.
• Experience with ECS and containerized services.
• Experience with GitHub Actions.
• Experience with Terraform and/or OpenTofu.
• Experience with CI/CD, Linux, networking, and production debugging.
• Strong observability skills across metrics, logs, and traces.
• Ability to write production-quality code or scripts in TypeScript and Bash.
• Ability to read and modify infrastructure, CI/CD, and application code.
• Good judgment around production risk, automation, permissions, and rollback.

Nice to have

• Kubernetes experience.
• Ruby experience.
• Experience building internal developer platforms or self-service infrastructure.
• Experience with coding agents, AI-assisted engineering workflows, repo-level agent instructions, evals, or agent guardrails.
• Experience improving incident response, deploy safety, or on-call quality.

What good looks like

After three to six months, we expect that:

• Deployments are less fragile and less dependent on manual intervention.
• Engineers are blocked less often by operational uncertainty.
• The production environment is easier to understand through service maps, runbooks, repo instructions, and observability.
• CI/CD gives faster and more trustworthy feedback.
• Agentic workflows are being used in practical, safe ways to help with debugging, deployment support, and operational improvements.
• We have clearer boundaries around what agents can inspect, suggest, and change.
• Production changes remain controlled, reviewable, auditable, and reversible.

What we do not want

This role is probably not right for someone who:

• Only wants to operate infrastructure manually.
• Treats AI agents as magic instead of unreliable junior collaborators with tool access.
• Optimizes for impressive demos over production safety.
• Cannot explain how they would constrain, observe, test, and roll back automated workflows.
• Thinks agentic engineering means removing humans from production decisions.
• Wants to bypass operational controls in the name of speed.

What you can expect at CopeCart

Hints for your application