Algon33 Blog

Algon

London, UK | algon33.ar@gmail.com

algon-33.github.io | lesswrong.com/users/algon | @Algon_33

PROFILE

Generalist project lead with a physics research background. I write well, synthesise complex technical material into accessible forms, manage small teams across diverse skill sets, and learn new domains fast. Two years leading an AI safety research communication organisation, producing several hundred reference articles, AI safety distillations, and research tools used across the field.

EXPERIENCE

Project Lead — AISafety.info (Stampy) Jul 2024 – Present

AISafety.info is a non-profit research communication project (founded by Rob Miles) producing living documents on the AI safety landscape, including technical research agendas, policy, and alignment proposals.

– Led a team of up to 10 contributors — writers, editors, programmers, designers, and researchers — with varying full-time, part-time, and volunteer involvement.

– Maintained and directed several hundred articles communicating AI safety concepts to a broad audience. Articles have been referenced by journalists, used in interview preparation by researchers, and serve as introductions to the field for new entrants to AI safety.

– Produced distillations of technical AI safety research, making frontier alignment work accessible to wider audiences.

– Oversaw development of an AI safety Q\&A chatbot, well-regarded by researchers in the field. Managed improvements to the RAG architecture and prompt engineering.

– Directed improvements to the Alignment Research Database, a comprehensive collection of AI safety literature cited in published research papers. Team built an MCP server enabling programmatic access for chatbots and research tools.

– Redesigned introductory article series; managed video content pipeline; handled analytics, infrastructure, and stakeholder communication.

AI Evaluation Specialist (Contractor) — Equistamp Jan 2024 – Jul 2024

– Designed and augmented AI capability evaluations for frontier model assessment.

– Sourced and validated high-quality evaluation benchmarks across diverse capability domains.

Freelance Research & Writing — Various Clients 2021 – 2024

– Atlas Fellowship: Authored 30-page research report on ambitious methods in education, analysing approaches that substantially outperform standard practice along axes of generality, effectiveness, cost, and scalability.

– AI Strategic Reasoning Assessment: Produced technical report on methods for detecting strategic reasoning capabilities in near-term AI systems, commissioned by an AI safety research organisation.

– Metaculus Automation: Built Python tool automating prediction submission for a top-ranked Metaculus superforecaster by extracting and replicating community probability distributions.

– Provided research advisory and strategic support to independent professionals on technical projects.

EDUCATION

MPhys Physics with Theoretical Physics — First Class Honours — University of Manchester 2015 – 2019

– Thesis: demonstrated that environmental noise can produce a constant-factor speedup in basic quantum computation operations (polaron transform applied to a Landau-Zener transition).

SELECTED OUTPUTS

– AISafety.info article library — Several hundred articles covering the AI safety research landscape, from introductory overviews to technical distillations. aisafety.info

– Alignment Research Database & MCP Server — Comprehensive database of AI safety literature, cited in published research. MCP server enables programmatic access for chatbots and research tools.

– AI Safety Q\&A Chatbot — RAG-based chatbot drawing on the full article library and database to answer questions about AI alignment research.

– Blog & writing — algon-33.github.io

KEY COMPETENCIES

Project & team management • Technical writing & distillation • AI safety research landscape • Research communication • Stakeholder management • Grant writing • Python • Data analysis • Physics & mathematical modelling