Contact · Sydney, Australia

Start with
the problem.

The most useful first message is the one that describes what you are trying to measure and why the current answer is not trusted. Everything else — stack, scope, timeline — follows from that.

Direct

Email is read properly rather than quickly. Expect a considered reply rather than an immediate one.

Sydney, Australia AEST / AEDT

Currently open to

  • Machine learning
  • Quantitative finance
  • Actuarial studies
  • Data infrastructure

Engagements

What tends to
work well.

  1. 01

    Applied machine learning

    Sensor fusion, multimodal alignment, retrieval-grounded generation, and the unglamorous parts underneath — data audits, canonicalisation layers, calibration and abstention policy. Particularly where labels are scarce and the honest architecture is not the fashionable one.

  2. 02

    Quantitative systems

    Strategy research with execution parity, risk engines, event-driven backtesting, and the operational safeties that matter more than any edge. Including the audit of an existing system that is not reproducing out of sample.

  3. 03

    Actuarial and health economics

    Cost and demand modelling, scenario engines, staffing and supply estimation, and the reporting layer that makes an uncertain figure legible to somebody who has to sign off on it.

  4. 04

    Research and open source

    Collaboration on published work, tooling for the agent ecosystem, and contributions to the projects listed in the corpus. Always open to a well-posed problem.

One line is enough

Tell me what you need
to be able to prove.