Building Explainers Random Thoughts - Drone 101
An experiment in interactive explainers: drone control, Fourier, Bret Victor, philosophical prompts, accessibility, and learning with LLMs.
Micheál Reilly
Some writing about things.
I am a software engineer writing about agentic AI tooling, learning systems,
institutions, and economic history. Most posts here are first-hand notes:
what I built, what broke, and what I would do differently. Every page is
also available as clean markdown, either by appending .md to
the URL or by requesting it with Accept: text/markdown. See
llms.txt for the curated index, or
about for more on what I work on.
Latest writing
Emerging Field of Epistemic Engineering or Stochastic Management
Practices around context, uncertainty, and evals are hardening into a discipline, with belief states at its centre: probabilistic, grounded views of a work process.
Cheaper Code Should Mean More Experiments
When working with agentic tools, the instinct to solve hard problems inside your main project is usually wrong. Isolate the core problem first, experiment freely, then bring what works back.
Mountain to Climb
A tool for visualising how long economic catch-up growth actually takes between any two countries or regions
Software Architecture That Agents Actually Like
Structural patterns that make codebases easier for AI agents to navigate: composition over monoliths, stateless mini-CLIs, typed languages, and the editor navigation test.
Using Agentic AI Tools for Anki Flashcard Creation
How Claude Code and similar tools can handle the tedious work of expanding concepts into multiple flashcards, and why Anki Connect is the key to making it work.
Freeport: A Lightweight Port Management Tool for Development
A thin wrapper around native commands to simplify port checking and process management during local development.
Exploring the Primitives of an Agentic Coding Setup
Reflecting on the foundational layers of agentic systems, from Git workflows and networking to storage and sandboxing, rather than high-level orchestration.
Notes on LLM Clock Face Limitations
LLMs struggle to generalise clock face generation beyond the standard 12 divisions, highlighting a specific gap in reasoning versus training data memorisation.
Notes on Robotics and Untapped Real World Potential
Exploring the adjacent possible in robotics, where falling hardware costs and advanced software create new opportunities for high-quality automation in niche domains like local sports broadcasting.