Contact sales

A paid AI research residency, in Melbourne

Late-stage PhD students and standout early-career researchers spend 3 to 6 months on their own question, with our GPU cluster, stack and team.

Apply with a proposal Our research

How it works

You own the project and set the direction.

You run the experiments on our GPU cluster, and we remove blockers and work through the problem with you.

Research areas

Six research areas open to residents.

Some projects go straight into Matilda, and others test ideas that may take years to reach a product.

  • Agents

    Make agents reliable when they act on live systems.

    Tool use · Planning · Memory · Multi-agent

  • Safety and alignment

    Find out when a capable system can be relied on, and why.

    Interpretability · Oversight · Evaluation

  • Training and algorithms

    Study how objectives and methods shape what a model learns.

    Objectives · Optimisation · Reasoning · RL

  • Data

    The data a model trains on sets what it can do.

    Curation · Synthesis · Mixture design

  • Multimodal

    Train models that reason over images, text and audio together.

    Vision · Language · Audio

  • Efficiency

    Make training and serving faster on our own GPUs.

    Training · Serving · Kernels

Who you are

Who the residency is for.

  • A late-stage PhD student, or a standout early-career researcher
  • Published at top venues, or clearly on track to
  • Comfortable running large-scale experiments and systems-level work
Length
3 to 6 months, paid
Where
On-site, hybrid, or remote with visits
Team
Day to day with the core team

Before you apply

Posts by the people you’d work with.

Company

What it's like being an AI Researcher on the Matilda team

We interview a lot of candidates who are making the move from academia into industry. It's a big transition and it makes sense that people carry assumptions from one world into the other. There's genuine overlap between the two, but the overlap is incomplete. The part that tends to surprise people most is how much engineering is involved. This post is specifically about what it means to be an AI researcher on the Matilda team, where engineering isn't a side activity but the bridge between havin

Dave Lemphers · 4 min read

Research

Same Dictionary, Different Geometry

Recent work shows Muon representations are more robust and transferable. We ask what that means anatomically: at 124M scale, firing-pattern matching says the sparse features are the same — Muon recovers the AdamW feature set up to seed noise, packaged into a hotter, sparser, more crowded, and more seed-reproducible residual-stream geometry.

Max · 8 min read

Apply

Send us your research proposal.

Tell us the question you want to work on, and why it matters.

Apply on Ashby Ask a question