Role details
Own the agent runtime behind a corporate travel platform that completes real bookings across web, email, Slack, Teams, WhatsApp and voice.
You'll work directly with the founder and engineering team to design and build the systems behind their AI agents. That means architectural decisions, production code, and solving the hard problem of making an agent act correctly when real money and real trips are on the line. This is a hands-on leadership role, not a pure advisory one.
The team works out of Melbourne or Sydney with flexible arrangements, so the day-to-day is built around getting things done rather than being seen to. The stack is Python, FastAPI, LangGraph and MCP/FastMCP on the backend, with PostgreSQL and Redis handling persistent state. LangSmith and Sentry give you proper observability into what the agents are actually doing in production. You don't need to have used every tool here, but you do need the engineering judgement to know when to trust the model and when to constrain it.
This is an early-stage company, which means the architecture is still being shaped and your decisions will carry real weight. You'll mentor the engineering team and set the technical direction, but you'll also be writing and reviewing code yourself. If you want a role where you hand off implementation to others, this probably isn't it. If you want to own something end-to-end at a company where the product is genuinely hard to build, it's worth a conversation.
What You'll Do
- Design and own the agent runtime, covering orchestration, tool execution, context management and persistent conversation state across all channels.
- Define the boundary between model reasoning and deterministic execution, with clear controls around bookings, payments and approvals.
- Build reliable integrations with travel distribution systems, airline and hotel APIs, payments and messaging platforms, including retries, idempotency and failure recovery.
- Build evaluation frameworks using conversation replays, regression tests and production traces, then use those results to drive release decisions.
What You'll Need
- A track record of shipping production AI agent systems with direct ownership of architecture, implementation and operational performance.
- Deep experience in agent orchestration including tool execution, durable workflows, state management and recovery across long-running tasks.
- Advanced Python skills and solid foundations in distributed systems, async processing, API design and database architecture.
- Experience building evaluation frameworks that measure task completion, tool-use accuracy and failure recovery, plus strong security judgement across tenant isolation, prompt injection and sensitive data handling.
About the Company
They're a corporate travel AI startup building infrastructure that lets travel management companies and their customers search, book and manage travel through natural conversation. Their agents handle real transactions across multiple channels, connecting to travel suppliers and business systems with the controls enterprises need.
Apply
Please click the 'Apply' button. Don't worry if your CV isn't up to date - just send what you have.