I've submitted my PhD thesis on modelling Transformers/LLMs as approximations of Solomonoff Induction, now I need a job. Lots of AI safety orgs offer a bonus for anyone referring a successful candidate - one place I've applied to offers $5,000. Refer me, claim the money and some play money!

Resolves YES if I accept a paid full-time position (not necessarily permanent) from someone referring me. The referrer isn't required to claim a referral bonus, but hey, why not try? (I've already applied to a few places, so you might not be able to claim them there; if you want to know if I've already applied somewhere, DM me)

Resolves NO if I don't accept a position or the offer was made without a referral.

About me

Name: Nathan "N.C." Young (seems worth restating that I am not @NathanpmYoung)

Location: Auckland, New Zealand. UK/NZ dual citizen, so able to relocate to and work in the UK without a visa. Willing to move elsewhere as well or work remotely

Qualifications: Submitted PhD thesis (currently under examination) on modelling Transformers/LLMs as approximations of Solomonoff Induction. Developed new formulation called Hierarchical Solomonoff Induction, replicated Learning Universal Predictors and Is SGD a Bayesian sampler? Well, almost to examine how well Transformers approximate Bayesian induction at training and test time separately.

Ideal role: Theory-first AI safety research scientist, developing a theoretical understanding of how AI works/should be expected to develop, supported with empirical observations/experiments.

Specific interest in unbounded models like Solomonoff Induction and AIXI, and algorithmic information theory in general. I got a tattoo of the AIXI definition last weekend, put it that way.

Main (public) research artifact: Hierarchical Solomonoff Induction: An Unbounded Machine Learning Model

Relevant links:

• https://nathanyoung.nz

• https://scholar.google.co.nz/citations?user=MYN0H7oAAAAJ

• https://github.com/NcyRocks

• https://calendly.com/ncyoung/

Availability: no current commitments, could start remotely immediately or overseas within a month or so

Additional qualifications:

• Good writer (technical and creative; have written plays for university theatre for 10 years, see https://www.scribd.com/user/479147549/Nathan-Young)

• Manifold moderator for ~3 years (but you probably knew that)

• Memorised pi to 115 decimal places

• If I hold my hands behind my back I can bring them to my front without letting go or stepping over

Hierarchical Solomonoff Induction: An Unbounded Machine Learning Model

Abstract page for arXiv paper 2608.01005: Hierarchical Solomonoff Induction: An Unbounded Machine Learning Model