About the Role: Johns Lyng Group has been helping Australians recover from disaster for more than 70 years and has grown to become Australia's largest integrated disaster recovery and building services provider, with operations across Australia and New Zealand.
Based in our Richmond office, you'll join a close-knit technology team working on a major AI project that will change the way our business operates. Rather than building proofs of concept, you'll develop AI solutions that solve real operational challenges, improve the way our people work and support a business processing thousands of jobs every year. If you're looking for the opportunity to build AI that will be used every day, this is it.
We Offer: - Opportunity to join a growing organisation with a strong pipeline of work
- Stability and job security with an established business
- The chance to make an impact across multiple business units
- Supportive, collaborative team environment
- Career progression with clear pathways
- Ongoing training, mentorship and development
- Discounts on technology, cars, fashion and leisure
- Modern systems and tools
Key Duties and Responsibilities: - Find routine, high-volume work across the business and turn it into AI-assisted workflows that people supervise rather than perform.
- Design, build and maintain custom AI agents, including tool definitions, control flow, state and memory, retries, escalation paths and stopping conditions.
- Integrate LLM and multimodal model APIs into production applications and build structured extraction from PDFs, scans, photos and emails with per-field confidence.
- Write and maintain evaluation sets from real production data, set confidence thresholds, and support staged rollouts backed by evidence.
- Design human-in-the-loop experiences that make AI confidence visible, simplify review and capture feedback for continuous improvement.
- Instrument AI systems with tracing, token usage, cost, latency and quality metrics, and build dashboards that provide meaningful insights.
- Build secure, production-ready AI solutions by managing prompt injection risks, permission boundaries, audit trails and approval gates.
- Build and maintain shared AI capabilities, including MCP servers, tool libraries, prompt and configuration versioning, AI gateways and evaluation tooling.
- Write clear documentation, build maintainable interfaces, and debug and support solutions through production and beyond.
Key Selection Criteria: - Proven experience shipping LLM-powered systems to real users and supporting them in production.
- Experience designing and building AI agents, including tool design, orchestration, state management and human-in-the-loop workflows.
- Strong Python and TypeScript skills, with experience building production-ready AI applications and services.
- Experience with prompt and context engineering, structured outputs, evaluation frameworks and confidence-based AI systems.
- Solid understanding of Model Context Protocol (MCP), AI security, prompt injection mitigation and responsible AI practices.
- Experience managing model selection, token usage, latency, cost and performance in production environments.
- Strong knowledge of leading AI models, including Google Gemini, Anthropic Claude and OpenAI GPT.
- Hands-on experience with Google Cloud, particularly Vertex AI, Gemini, Cloud Run, BigQuery and related services.
- 5+ years' experience building and shipping production software, including API design, event-driven architecture, CI/CD, infrastructure as code, Git and testing.
- Experience working across existing technology stacks, with the ability to quickly understand and contribute to established codebases.
- Experience with vision and multimodal models, document extraction, AI observability, retrieval, vector search or workflow engines will be highly regarded.
Our environment: - Cloud: Google Cloud is the primary platform and where most new AI work lands. Microsoft Azure is also in the estate. Both are in play.
- AI: Agent Platform and Gemini, Python agents deployed as containerised services, Microsoft 365 and Graph integration, with third-party model providers used where they are the better fit.
- Applications: Service-based .NET platforms with workflow engines and Angular front ends, alongside long-standing line-of-business systems carrying significant document volume and many inbound integrations.
- Data: SQL Server, PostgreSQL, BigQuery, and enterprise BI tooling.
- Engineering: Azure DevOps for source control, pipelines and boards. Terraform for infrastructure. Datadog for observability.
Please note: A written technical assessment, criminal background check and medical will form part of the recruitment process.
Agency Notice: JLG do not accept unsolicited resumes or candidate profiles from recruitment agencies. Please do not forward resumes to hiring managers, or any other company employees. JLG are not responsible for any fees related to unsolicited resumes.