Experience

2026 – Present
Technical Co-founder
Dokimasia
2023 – 2025
Research Scientist
FAR AI
2022 – 2023
Member of Technical Staff
Redwood Research
2021
Summer Research Fellow
Center on Long-Term Risk
2019
Research Intern
Microsoft Research Cambridge

Education

2017 – 2021
PhD Machine Learning
University of Cambridge
2016 – 2017
MSc Computer Science
University of Oxford · Distinction
2012 – 2016
BSc Computer Science
Pompeu Fabra University · 1st in class

Awards

2017
Malmö AI Challenge Winner
1st & 3rd place · $20k Azure credits
2016
la Caixa Fellowship
6.6% acceptance rate
NeurIPS 2023 · Spotlight
Towards Automated Circuit Discovery for Mechanistic Interpretability
A. Conmy, A. Mavor-Parker, A. Lynch, S. Heimersheim, A. Garriga-Alonso
~400 citations arXiv:2304.14997
ICLR 2019
Deep Convolutional Networks as Shallow Gaussian Processes
A. Garriga-Alonso, L. Aitchison, C.E. Rasmussen
~330 citations arXiv:1810.05148
Alignment Forum 2022
Causal Scrubbing: A Method for Rigorously Testing Interpretability Hypotheses
L. Chan, A. Garriga-Alonso, N. Goldowsky-Dill, R. Greenblatt, et al.
~90 citations
2025
Open Problems in Mechanistic Interpretability
L. Sharkey, B. Chughtai, [...], A. Garriga-Alonso, et al.
~100 citations
Machine Learning Why Deep Learning Works: Specificity, Not Flexibility

The common narrative that deep learning works because neural networks are flexible universal approximators misses the point entirely...

Updated Jan 15 12 min
Philosophy On Consciousness and Moral Weight

How should we think about the moral status of potentially conscious AI systems? A framework for uncertainty...

Updated Dec 20 8 min
Mathematics Testing Integrals: A Practical Guide

How to verify your integral computations are correct using randomized testing and numerical methods...

Updated Nov 8 5 min
Ethics On Killing vs Letting Die

The traditional distinction between action and inaction in moral philosophy deserves re-examination...

Updated Oct 14 7 min
Tools Remote Development with Unison

A practical guide to using Unison for seamless local-remote file synchronization during development...

Updated Sep 22 6 min
Philosophy Alternative Population Ethics

Exploring alternatives to total and average utilitarianism that avoid the repugnant conclusion...

Updated Aug 5 10 min