Applied AI & LLM Engineering
Engineers who ship LLM-backed systems to production — retrieval, agents, evals, fine-tuning — not research for its own sake.
A$160k – A$280k package
This is the fastest-growing AI function in Australia and the one most often mis-scoped. The role most companies need in 2026 is not a research scientist — it is an engineer who can take a frontier model, wrap it in retrieval, tools and guardrails, build the evaluation harness that proves it works, and run it in production within a real latency and cost budget.
The failure mode we see weekly: a job spec asking for a PhD, publications and "deep transformer internals" for what is actually an applied engineering role — then losing every good applicant to a six-week process. The candidates who excel here look different: strong software engineers with genuine LLM production scars — people who can talk concretely about eval design, context management, failure taxonomies and cost curves, because they have shipped through them.
We assess for shipped systems, not keyword familiarity. Every shortlisted candidate has been interviewed by a principal against a role-specific rubric, and each shortlist ships with a transcript-cited technical evidence pack, so your engineers can see exactly what was asked and answered before the first interview.
- AI Engineer / LLM Engineer
- Applied ML Engineer
- Retrieval & agents (RAG, tool-use, orchestration)
- Evaluation & fine-tuning specialists
- Founding AI Engineer
Where this hiring goes wrong
Applied AI engineer ≠ research scientist
One optimises a product under constraints; the other advances capability. Hiring the second to do the first costs you 30–40% more compensation and usually ends within a year — in either direction, the mismatch is expensive.
LLM engineering ≠ data science
The skill core has moved: systems engineering, evaluation design and product judgement now matter more than classical modelling. A strong data scientist is not automatically a strong LLM engineer, and vice versa.