A sci-fi mind. An engineer’s hands.
I work on machine learning and decision systems. My background is in computational astrophysics; my current interests are learning, human judgement and how people work together.
A question I keep returning to
“What helps people make better decisions, learn well and work together?”
I began in computational astrophysics, studying stellar dynamics and binary black holes through numerical simulation and statistical modelling.
I then worked on machine learning for consumer products, marketplaces and advertising, where a model has to be useful within the constraints of a working system.
Building a company brought me closer to the people using the technology. It made me more attentive to what they need, and to the gap between a promising idea and something useful.
A way to browse the questions behind my work. The larger nodes represent areas of interest; the smaller ones link to essays, research and project notes.
Select a topic or a piece of writing to see how it relates to the other work collected here.
An open research ledger testing which LLM-mediated learning policies produce durable, transferable capability per unit of learner effort outside the tutoring session.
The study asks whether making a design judgement before seeing an expert solution improves later recall and transfer compared with explanation-first tutoring.
The subject matter has changed. The habit of making a model, testing it and revising it has stayed with me.
Ph.D. Physics (2013–2018) · LSST Data Science Fellow
I began in computational astrophysics, studying stellar dynamics and binary black holes through numerical simulation and statistical modelling.
Applied ML & Decision Intelligence (2018–2024)
I then worked on machine learning for consumer products, marketplaces and advertising, where a model has to be useful within the constraints of a working system.
Independent projects & ventures (2018–present)
Building a company brought me closer to the people using the technology. It made me more attentive to what they need, and to the gap between a promising idea and something useful.
Good work depends on more than individual ability. Shared context, trust, complementary skills and sensible incentives matter, as does room for people to exercise their own judgement.
I judge technology by what it helps people do: understand a problem, make a choice, learn a skill or spend their time better. Technical elegance is valuable when it serves that purpose.
I try to make assumptions explicit, build something that can be tested and pay attention to the result. Revising an explanation is part of the work.
* These are ongoing questions. The linked work distinguishes observations, proposals and research findings.
Research notes, software and design specifications at different stages of development. The linked repositories describe the work and how it is evaluated.
Studying what people retain after learning with AI.
An open research ledger comparing approaches to learning with language models. The focus is on what learners retain and can apply independently, relative to the effort involved.
Specifications for a household agent that learns from spoken feedback.
Five subsystem specifications covering perception, speech recognition, memory, adaptation and evaluation. The design targets local hardware costing under $200, with signed evaluation records.
Access to AI tools through an ordinary message.
An experiment in using iMessage to reach internet-connected agents when normal browsing is limited. It explores how a familiar, asynchronous interface can work with constrained connections.
Technology and community projects with older adults.
Work on AI companions, accessible interfaces and community activities in Palo Alto, informed by older adults’ experiences of technology, mobility and learning.
Separating real-time inference from asynchronous optimisation.
An architecture proposal for an eight-agent control plane alongside a latency-sensitive serving path. Accompanying notes consider advertising economics, experimentation and causal measurement.
Notes on governance, cooperation and possible futures.
An evolving collection of proposals about social contracts, liquid democracy, local manufacturing and ways of turning ideas into practical experiments.
Collected from Domi’s Mind Waves, Medium, GitHub and the research archive. Each entry links to its source.
An open research ledger testing which LLM-mediated learning policies produce durable, transferable capability per unit of learner effort outside the tutoring session.
Five subsystem specifications for perception, speech, memory, adaptation and evaluation, with a hardware cost target below $200.
An experiment in using messaging to reach internet-connected agents over limited connections, including in-flight Wi-Fi and satellite text.
Examines Jane McGonigal's game mechanics (goals, rules, feedback systems, voluntary participation) through conversations with senior learners and retirees.
Observations from work with older adults at Kin Domi, alongside six proposals for more accessible driving-assistance interfaces.
A reflection on team alignment, complementary skills and sustainable working habits.
Notes on the eighth revision of a working model built around shared calendars, asynchronous collaboration and weekly reflection.
An architecture proposal for an eight-agent control plane and an ad-serving path targeting latency below 50ms at p99, with notes on performance advertising economics.
A personal reflection on connections between astrophysics, Confucian ethics and work across disciplines.
Notes on customer research, community outreach and forming the team behind Kin Domi, Inc.
Examines the mechanics of coaching, intellectual honesty, and why shared context determines the effectiveness of feedback.
An architectural discussion of personal AI memory, local voice processing, privacy in query rewriting and browser context.
Uses the visual ergonomics of dark mode as a case study to critique short-term agile feature shipping versus long-term user welfare.
An analysis of behavioural data, scaling and latency constraints in advertising, with questions about the costs of continued growth.
Proposes richer intent models, compact user profiles and interfaces that let users request different advertisements.
Explores how traffic-light timing might be inferred from crowdsourced trajectories, alongside reflections from Wolfram Summer School.
An exploration of documentation bottlenecks, the bandwidth of speech and writing, and institutional information barriers.
A 2023 essay on language models, search economics, symbolic computation and the limitations of feedback used for alignment.
Explores information asymmetry, resource coordination and proposals for liquid democracy.
Doctoral work on globular clusters and binary black holes, alongside notes on reproducible computational research.
If you are working on a related question, I would be glad to compare notes.
Questions, useful references and considered disagreements are welcome.
