Behavioural Researcher

Maincode is an AI research and engineering company. We build Matilda, Australia's large language model, trained, served and deployed here in Melbourne. We run our own GPUs in local data centres and we build everything above them: the racks, the training and serving systems, and the products that run on them.

The hardest problem at Maincode is not the tech. It is finding the small number of people who can build a frontier AI company. You will join the talent team at Maincode and own a piece of that problem.

Your job is to understand people. Personality, motivation, belief, mindset: what someone is actually made of, and whether it is what a team like this needs. You will make reads on real candidates, and you will rebuild the way we make them: the interviews, the instruments, and the agentic systems the pipeline runs on. Nothing about our process is sacred. If a stage does not predict, you kill it and build something that does.

Maincode is built on a small number of obsessive people working on hard problems. We are direct with each other, the bar does not move, and the team is the product, so the calls we make about people are the most important calls we make. You should recognise that way of working and want more of it.

You do not need a background in talent. If anything, we prefer that you do not have one.

Read candidates: their work, their words, their rooms; produce the judgments the team hires from.

Design and test the instruments: interview formats, screens, rubrics. Treat the pipeline as a running experiment and follow the results.

Build with AI. The pipeline is agentic end to end. You extend it, not just operate it.

Work day to day with talent and the founders, and calibrate your reads against how people actually turn out.

You are deeply fascinated by understanding human behaviour. You may have studied psychology, or not. You read well. You understand, far better than others, why someone does what they do. You understand and build systems. You hold a high bar. You are comfortable learning deeply technical things.