The opportunity
Build agents that do useful work across connected systems. You will design tool contracts, state transitions and approval points for workflows such as enquiries, internal operations and business assistance.
The focus is dependable behaviour rather than a convincing demo. You will test failures, preserve auditability and make it clear what an agent has actually completed.
What you’ll do
- Develop tool-using agents with explicit permissions and boundaries
- Implement durable workflow state and recovery paths
- Connect business APIs with validated tool inputs and outputs
- Build human review and handoff steps for consequential actions
- Evaluate multi-step task completion and failure behaviour
- Add traces and operational documentation for support teams
What you’ll bring
- Strong backend engineering skills in Python or TypeScript
- Experience with LLM tool calling and agent workflows
- Understanding of authentication, API contracts and idempotency
- Ability to test multi-step systems and ambiguous outcomes
- Clear thinking about state, retries and human approvals
Nice to have
- Experience with workflow engines or job queues
- Familiarity with MCP integrations
- Experience building CRM or operations integrations
Your toolkit
Make your next move.
Share your CV, portfolio or relevant project links through the application form. Include the role title and job reference in your response.
Salary and role-specific working arrangements will be discussed during hiring.
Apply for AI Agent Engineer. Include the job reference ADZ-003 in your response.
Open application form ↗Applications are handled separately from this public job board. Privacy policy.
