Machine Learning Engineer (Agentic AI)

Job description

Our Purpose Planday from Xero is a leading digital solution that uncomplicates everyday scheduling and workforce management by making it easier for businesses and shift workers around the world to communicate, collaborate, and get work done. Powered by a community of local industry experts, Planday provides a best-in-class digital platform that is easy to use, accurate, secure, and compliant with local needs and standards. From payroll and accounting to POS and reporting, its open API and tech ecosystem is scalable to fit shifting business needs and to build an engaged, flexible workforce. Founded in 2004, Planday is headquartered in Copenhagen, Denmark and supports over 400,000 users across Europe, Australia and the US. Planday was acquired by Xero in 2021. How you’ll make an impact As a Machine Learning Engineer for Agentic AI & ML, you will work at the intersection of research and production. You will collaborate closely with internal stakeholders and customers to translate ambiguous challenges into effective, shipped features. You will work on features throughout the lifecycle - from discovery and model building to production deployment and maintenance. By designing and shipping ML and agentic systems, you will expand the platform's capabilities to fundamentally reshape how business owners run their operations and how workers manage transparency and work-life balance. What you’ll do Build and optimize ML models to enhance and extend Planday, framing problems from real metrics and engineering robust feature pipelines. Design and ship agentic systems involving tool use, function calling, and multi-agent orchestration for reliable planning workflows. Deploy and operate model services in production, ensuring performance for cost and latency through caching, batching, and quantization. Design and run rigorous evaluations, including LLM-as-judge and regression suites, so quality changes are always measured. Collaborate on turning research prototypes into reliable, observable production workflows and pipelines. Success looks like Stable and sustainable support for live ML services in production as our variety of deployments grows. Successful release of new features like auto-scheduler models and intelligent AI agent needs. Reduced ambiguity in model quality through the implementation of repeatable and high-fidelity evaluation sets. Proactive maintenance and de…