AI Platform Engineer

We use Claude Code, but local environments, agent configurations and access to development tools vary across the team. We need a platform engineer to build and own a consistent foundation for AI-assisted development. You’ll work with our technical lead to assess the current workflow, design the platform and implement a pilot across two services before expanding adoption. Your scope Sandboxed environments: provide reproducible, isolated environments for developers and coding agents, with dependencies, test services, controlled network access and support for parallel tasks. Access and security: implement short-lived credentials, scoped permissions, secrets isolation and approval boundaries for infrastructure changes. CI/CD: standardize build, test and deployment workflows. Integrate agent-generated changes with automated validation, security checks, infrastructure previews and human review. Temporary environments: automate provisioning and cleanup of environments for validating pull requests, including isolated test data, access controls and cost limits. Shared agent capabilities: establish repository guidance, reusable development workflows and authenticated access to approved documentation and engineering systems. Platform operations: monitor environment provisioning, pipeline reliability, agent activity and usage costs. Define ownership, operational procedures and onboarding documentation. Our application consists of TypeScript services and Python AI workers running on AWS. You’ll coordinate with our cloud platform engineer on infrastructure and access, while owning the development environments and automation used by the engineering team. We’re looking for strong production experience in AWS, Infrastructure as Code, CI/CD, workload isolation and developer platforms. You should also have introduced AI coding agents into a team and understand how to enforce permissions and execution controls. Success means a developer can start an isolated workspace, complete a task with Claude Code, open a tested pull request and validate it in a temporary environment with minimal manual setup. We’ll measure provisioning time, pipeline reliability, reviewer rework and operating costs. Initial contract: 6–8 weeks, 20–25 hours per week, with European working-hour overlap. The initial scope is a pilot across two services.