DevOps / AgentOps Engineer, GTM Systems

Start free →

Or apply yourself on the official page →

About the role

About Anthropic Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems. About the role Go To Market ("GTM") Systems builds and runs the platform that Sales and contracting operate on, a production Salesforce org plus a growing portfolio of custom internal applications and agents. These systems process revenue-impacting transactions, guide critical business processes, and produce data the rest of the company runs on. Today the path from request, development, test, and pull request to production push is manual and slows the team down rather than speeding it up. We are hiring a DevOps / AgentOps engineer to build and run that path. You will stand up and maintain an AI-first CI/CD pipeline covering both Salesforce and custom application code. This release pipeline will support three modes: fully agentic, human-in-the-loop, and AI-assisted. Anthropic is an unusual place to do this. Our own frontier models are part of the toolkit and the question we want you to answer is what a delivery system looks like when you design it for agents and humans working side by side rather than retrofitting agents into a pipeline built for humans alone. You will work closely with a team of business systems analysts, software engineers, and integration engineers to develop the solution to accelerate their everyday work. The pipeline you build is the one path to production for a system under audit, so the guardrails, approvals, and audit trail are part of the product, not an afterthought. Key responsibilities • Design, build, and operate the CI/CD pipeline for a production Salesforce org and a portfolio of custom internal applications, from branching and validation through deploy and rollback • Define how a change is classified and routed into an agentic, human-in-the-loop, or AI-assisted lane, what gates each lane passes through, and how a change escalates between lanes when validation fails or risk changes • Build the agent operations layer: what agents are scoped to do in the pipeline, what tools and permissions they hold, how they are sandboxed differently for Salesforce and custom-app work, and what they are not trusted to do unsupervised • Implement approval workflows that keep PR review and business sign-off as distinct gates, with clear rules for which changes need which, who is asked, and what evidence they see • Own environment strategy, secrets handling, deploy observability, and kill-switch and rollback mechanics across all three lanes • Produce the audit trail and evidence a SOX-scoped platform needs • Measure and report on the pipeline itself: lead time, failure rate, share of work moving through each lane, and where humans are still doing work an agent should be doing • Pair with engineers on the team as they move onto the pipeline, and turn what you learn from them into better defaults Minimum qualifications • Have built and run a CI/CD pipeline for a team, including the branching model, merge gating, and rollback path, and can explain the choices you made and what you would change • Have hands-on experience with Salesforce deployment mechanics (metadata API, source-tracked development, org-to-org promotion) or a comparable enterprise platform • Work fluently with AI coding tools and agentic workflows, and have opinions about which kinds of engineering work belong in the hands of an agent, which need a human review, and where the boundaries are • Design for the failure case: you think about blast radius, sandboxing, permissions, and what happens when an automat

View the official posting →