Experience

Epoch

• Research Services
• 1 - 100 Employee

• Research Mentee

  • Jul 2023 - Present

  I'm analyzing major paradigm shifts throughout the history of AI in order to forecast how often we can expect to see them in the coming years. My research is supervised by Pablo Villalobos in a joint program with Epoch and the Forecasting Research Institute. I'm analyzing major paradigm shifts throughout the history of AI in order to forecast how often we can expect to see them in the coming years. My research is supervised by Pablo Villalobos in a joint program with Epoch and the Forecasting Research Institute.



University of Bristol

• United Kingdom
• Higher Education
• 700 & Above Employee

• Research Intern

  • Jul 2023 - Present

  I am working on mechanistic interpretability of LLMs under the supervision of Dr. Conor Houghton, specifically analyzing how GPTs "understand" the difference between verbs and nouns. I am working on mechanistic interpretability of LLMs under the supervision of Dr. Conor Houghton, specifically analyzing how GPTs "understand" the difference between verbs and nouns.



Alignment Research Engineer Accelerator (ARENA)

• London, England, United Kingdom

• Research Participant

  • May 2023 - Jun 2023

  High-intensity 6-week upskilling and research program in the SERI MATS offices in London. I primarily worked on interpreting the “goals” of GPT models trained to solve gridworld environments. I also worked on merging ACDC with attribution patching (preliminary results increased its performance by about 20,000% and won us #1 Prize in an ApartAI hackathon), interpreting OthelloGPT, and replicating the core findings of the IOI paper. High-intensity 6-week upskilling and research program in the SERI MATS offices in London. I primarily worked on interpreting the “goals” of GPT models trained to solve gridworld environments. I also worked on merging ACDC with attribution patching (preliminary results increased its performance by about 20,000% and won us #1 Prize in an ApartAI hackathon), interpreting OthelloGPT, and replicating the core findings of the IOI paper.



AI Safety Camp

• Research Contributor

  • Feb 2023 - Jun 2023

  My team of 5 researchers is exploring the connection between human brain circuitry and artificial neural networks. The goal is to use fMRI data to induce human-like biases in a large language model and make it score higher on a moral reasoning dataset. My team of 5 researchers is exploring the connection between human brain circuitry and artificial neural networks. The goal is to use fMRI data to induce human-like biases in a large language model and make it score higher on a moral reasoning dataset.



Longitude 103

• United States
• Software Development

• Senior Full Stack Developer

  • Feb 2020 - Jun 2023

  I'm the technical lead on most of our company’s major products. I handle everything from communicating with our customers about bugs and feature requests to designing the software architecture, implementing the front end and back end, and deploying everything.

• Full Stack Developer

  • Aug 2019 - Feb 2020



Živnostník

• Independent Researcher

  • Jan 2023 - May 2023

  I'm focusing on solving the inner alignment problem - currently exploring ideas around reward modeling/inverse reinforcement learning and neuroconnectionism. I'm collaborating on some of my research with scholars at the Future of Humanity Institute (University of Oxford) and the Center for Human-Compatible AI (UC Berkeley) I'm focusing on solving the inner alignment problem - currently exploring ideas around reward modeling/inverse reinforcement learning and neuroconnectionism. I'm collaborating on some of my research with scholars at the Future of Humanity Institute (University of Oxford) and the Center for Human-Compatible AI (UC Berkeley)