Go-live is where the AI bill starts, not where the project ends.
FERG Search is a specialist recruiter for MLOps and LLMOps engineers in Australia and New Zealand, permanent and contract: the engineers who deploy, evaluate, monitor and control the cost of models and agents on Microsoft Foundry and Azure Machine Learning. Every search is run by the founder, Fintan Ferguson.
Most AI budgets are written for the build. The cost that surprises people arrives afterwards: a model version changes, answers drift, usage triples, and nobody can say whether the agent is better or worse than it was last month. MLOps and LLMOps engineers are the people who can, and in the programmes we see the seat is usually created after the first incident, not before.
First shortlist within 48 hours of an agreed brief.
The people who measure AI after go-live.
- MLOps engineers on Azure Machine Learning and Microsoft Foundry.
- LLMOps engineers who run evaluation, tracing, and prompt and model versioning.
- AI platform engineers who own the deployment pipeline for models and agents.
- AI reliability and observability engineers.
- Engineers who own AI capacity, token spend and provisioned throughput.
Checked technically and commercially, before you meet them.
- The evaluation set they built, and how often it runs.
- What happened the last time a model version changed underneath them.
- The deployment pipeline, and how a bad release is rolled back.
- The monthly cost line they own, and what they cut.
What the seat pays.
MLOps briefs stall when the seat is budgeted as DevOps support and the job is owning model risk.
MLOps and LLMOps bands are not in the first edition of the FERG Search Salary Guide. The nearest published figure is the senior AI engineer band: $170k to $210k base in Sydney, excluding super. Until then, ask: we will give you the band we are seeing accepted, declined and countered for the seat you are briefing.
Much of the bench is already in your building.
Most of the MLOps and LLMOps engineers we meet came from DevOps, data engineering or platform teams, not from research. They already know pipelines, monitoring and cost. The step is learning how models and agents fail. That makes your cloud and data teams the first place to look, and the open market for people who have already made the step is small.
Straight answers on MLOps and LLMOps.
What is the difference between MLOps and LLMOps?
MLOps covers the lifecycle of machine learning models you train: data, training, deployment and monitoring. LLMOps covers the large language models and agents you mostly do not train: evaluation, prompt and model versioning, safety checks, tracing and token cost. Most agent programmes we see need more of the second than the first.
Who recruits MLOps and LLMOps engineers in Australia and New Zealand?
FERG Search recruits MLOps, LLMOps and AI platform engineers on Azure for Microsoft partners and end users across Australia and New Zealand, permanent and contract. The founder, Fintan Ferguson, runs every search from Melbourne.
When should we hire an LLMOps engineer?
Before the second agent goes live. One agent can be watched by the people who built it. Several agents, on models that change underneath them, need someone whose job is measuring them.
Can a DevOps engineer move into MLOps?
Often, and it is the most common route we see. Pipelines, environments and monitoring carry straight over. The new part is evaluation: deciding what a good answer is and testing for it on every release.
Brief an MLOps or LLMOps role.
Tell us how many models and agents are live, and who would notice if one got worse.