Senior Site Reliability Engineer, Agentic… at SewerAI

About SewerAI Corporation

SewerAI is transforming underground infrastructure management through AI-powered inspection and risk analysis. Our platform helps contractors, engineering firms, and utilities unlock valuable insights from sewer inspection data—turning hours of manual video review into actionable intelligence in minutes. After doubling our customer base over the past year, we’re now entering an exciting phase of accelerated growth.

About the Role

We're hiring a senior engineer to bring AI agents into how the whole company works. You'll find the repetitive, manual work that doesn’t need to exist, in engineering and well beyond it, and build agentic workflows that take that work off people's plates and let them do things they couldn't do before. You'll also own the internal platform those agents run on, so they're reliable, observable, and secure.

This is a new kind of role, and it doesn't fit a standard title. It sits between SRE, automation, and applied AI. You don't need to be a career software engineer. You do need to think like an operator, write solid code, and have good judgment about where an agent helps and where it doesn't. You'll work with people in every part of the company, so learning how each team works matters as much as the technology.

This is a remote role on the SRE team. It is not part of the on-call rotation.

This is a US-based position. Candidates must reside in the US.

What You'll Work On

Find the Work Worth Automating: Work with engineering, operations, customer success, finance, and other teams to find high-value automation opportunities, then design, build, and ship agentic workflows for them.

Eliminate Engineering Toil: Automate SRE and engineering chores such as alert triage, ticket creation and routing, incident follow-ups, dependency updates, and CI maintenance.

Build for Non-Technical Teams: Ship tools that teams can rely on without needing to understand how they work under the hood.

Own the Agent Platform: Run the internal platform our agents run on, including sandboxed execution, secrets management, network egress controls, permissions, and cost controls.

Treat Agents Like Production Systems: Back every workflow with evals, observability, alerting, runbooks, and clear failure modes.

Set the Guardrails: Define what agents can access and do, with least privilege, human approval for risky actions, and audit trails.

Make It Repeatable: Document patterns and build reusable components so others can safely build their own agent workflows.

Prove the Value: Measure impact (hours saved, response times, adoption) and retire workflows that don't pay off.

Required Technical Skills

Agentic Workflows: Hands-on experience building and running agentic or LLM-powered workflows in a real environment, not just prototypes.

Programming: Strong Python or TypeScript scripting, including API integrations and glue code, and the ability to read and modify production code.

Cloud Infrastructure: Solid working knowledge of AWS and cloud security fundamentals.

Infrastructure as Code: Experience with Terraform/OpenTofu or a similar tool.

CI/CD: Experience building and maintaining automated pipelines, ideally GitHub Actions.

Nice-to-Have Skills

Security Background (big plus): Experience with agent and LLM risks (prompt injection, data exfiltration, over-privileged tools), secrets management, IAM, or sandboxing and isolation.

Business Knowledge: Familiarity with how teams like sales, finance, customer success, or operations work day to day. This isn't required. How quickly you can learn a new domain matters more than what you already know.

Workflow orchestration with Temporal.io or similar.

LLM evaluation frameworks and observability tooling.

Experience with Claude Code, MCP, and integrations with tools like Slack, Linear, and Notion.

Experience in the infrastructure, utility, or geospatial domains.

What We're Looking For

6+ years in SRE, DevOps, infrastructure, or platform engineering.

An operator's mindset: you think about what happens when a workflow fails at 2am, before it ships.