Description

Lead Engineer - AI Platform & DevOps
Contract: Initial engagement of roughly 3-4 months, tied to the current build phase, with a strong chance of rolling into a larger rollout afterward
Location: Remote, but Australia-based only

Start: Immediate

The engagement

We are working with a private health insurer in Australia is having an AI-assisted software delivery platform built for them from scratch. The idea: AI agents handle pieces of the business analysis, build and review work inside the software delivery pipeline, but nothing consequential happens without a human sign-off. GitHub kicks off the work, AWS's event infrastructure routes it, Bedrock-hosted agents execute it, and everything is anchored to a semantic layer owned by the client that keeps every agent working from the same organisational context. Standing up that semantic layer is also part of this build.

What you'd actually be doing

You're the person turning this into working software, infrastructure, delivery pipelines, the agent runtime and the semantic layer, following direction from the engagement's architects, but bringing real opinions shaped by what you've actually built before. You'll also be expected to pass knowledge across to the client's engineers as the build progresses, not just leave documentation behind.

This isn't a role for someone who only wants to do AI work. Because it's starting from zero, the first stretch is heavy on core platform engineering, Terraform, AWS, identity and access, CI/CD and the AI/agent layer sits on top of that once the foundation's in place. If you've worked this exact combination before, solid infrastructure background with genuine AI/agent exposure, you'll be in your element. If you're purely AI-native without the platform depth, this probably isn't the right fit.

Tech environment

• Cloud: AWS, event-driven serverless architecture, multi-account setup with tight governance
• AI/Agents: Bedrock and Bedrock AgentCore running Claude models, with guardrails, evaluation and observability wired in
• Retrieval/Knowledge: a mix of vector and graph retrieval underpinning a client-owned semantic layer
• Delivery: Terraform, GitHub and GitHub Actions, secure keyless CI/CD, a locked-down container supply chain
• Governance: policy-as-code, human sign-off gates built into GitHub, full audit trail end to end
• Languages: Python and/or TypeScript; Java experience is a bonus

What they're looking for

Must have:

• 7+ years in software or platform engineering, with recent time at a senior or lead level on teams that actually ship
• Strong AWS background, especially event-driven serverless patterns and the security/governance side
• Genuine production Terraform experience across multiple accounts and environments, run through CI rather than by hand
• Solid DevSecOps/CI-CD grounding, GitHub Actions, keyless cloud authentication, container supply-chain security, promotion gates, rollback processes
• Has taken LLM or agent-based systems into production, or close to it, including the guardrails, evaluation and observability that go with it
• Understands retrieval engineering properly, embeddings, vector indexes, and where retrieval quality typically falls apart
• Strong Python and/or TypeScript

Nice to have:

• Semantic layer or knowledge platform experience ontologies, entity resolution, combined vector-and-graph retrieval, content managed through Git-based review.
• Direct experience with Bedrock AgentCore, MCP tooling, or agent SDKs
• Policy-as-code background
• Delivery experience in regulated Australian settings, comfortable with data-sovereignty requirements
• FinOps experience specific to AI workloads
• Has worked in consulting or embedded directly inside a client's team before