We’re building model-orchestrated AI agents on AWS to power our operations — a governance layer on top, domain agents underneath, each running scoped “skill” prompts against a shared toolset.
Some workflows run as full automations; others are human-in-the-loop, where the agent drafts and a person reviews before anything executes.
We’re looking for someone who can start immediately and move fast without sacrificing accuracy.
What You’ll Do
Build and deploy agents/workflows in our AWS environment
Author agent skills (prompt + scoped tools) for each domain
Build both fully automated workflows and human-in-the-loop tasks:
agent drafts → human review/approval → execute
Wire tool integrations and guardrails, including scoped access, review gates, and kill-switches
Ship clean, documented, testable builds through our CI pipeline
Operate and coordinate multiple AI coding/agent sessions simultaneously to accelerate delivery
Break larger projects into parallel workstreams, delegate those workstreams to agents, review their output, resolve conflicts,
and integrate the final build
Must Have
Strong AWS experience:
ECS Fargate
Lambda + SQS
DynamoDB
API Gateway
Cognito
CodePipeline
Amazon Bedrock experience, or comparable LLM/agent framework experience
Proven experience building LLM agents or tool-using workflows in production
Experience building integrations against third-party APIs, including REST APIs and webhooks
Reliable overlap with US working hours (Eastern to Pacific) for huddles and reviews
Ability to start immediately
Strong judgment, testing discipline, and attention to detail
Critical: Multi-Agent Execution
We are specifically looking for someone who knows how to use AI development tools as a force multiplier, not someone who launches one Claude Code session and waits for it to finish.
You will be provided with multiple Claude Code accounts/environments and are expected to:
Run multiple agents and development workstreams in parallel
Give each