Robotics, AI systems, and public-data research.
I like building things where the claim has to survive contact with the real world.
I'm a Computer Science student at the University of Edinburgh working across low-cost robotics, AI interfaces, and empirical work on AI's labour-market effects. Recent projects include a sensorless force-feedback teleoperation system, a realtime speech-to-whiteboard app, and a public-data paper on young workers falling behind in AI-exposed occupations.
I co-founded EdinburghAI, worked as an AI Engineer Intern at V7 Labs, and keep getting pulled toward projects that are a little too ambitious for the time available. That is usually where the interesting bits are.
| Project | What it is | Links |
|---|---|---|
| Sensorless Bilateral Teleoperation | A low-cost dual-arm teleoperation system with custom 2x-scaled 3D-printed follower arms and force feedback using only motor current. | Repo / Project page / DOI |
| Realtime Speech-to-Whiteboard | A browser app that turns spoken architecture descriptions into editable tldraw diagrams through the OpenAI Realtime API. | Repo / Demo |
| EdinburghAI Workshops | Practical machine-learning workshop materials for students learning the basics through applied projects. | Repo / EdinburghAI |
| The Missing First Rung | A public-data economics paper testing whether AI-exposed occupations are tilting away from young workers before headline unemployment moves. | Paper / Interactive essay / DOI |
The current robotics obsession is cheap bilateral teleoperation: commodity leader arms, custom printed follower links, a ROS 2 stack, and motor-current signals pushed just far enough to become haptic feedback.
The system works well enough for users to feel contact, but it also exposes the hard limitation: motor current alone cannot reliably separate gravity load from object contact. That tension is the interesting part.
- AI agents and interfaces that are useful immediately, not only as demos.
- Student tools, including EdinburghAI workshop materials and a University of Edinburgh grade calculator.
- Hackathon/product builds across voice interfaces, AI radio, prediction markets, dashboards, and marketing agents.
- Writing about AI, labour markets, robotics, and the transition details that are easy to miss.
Python, TypeScript, React, ROS 2, LaTeX, Jupyter, FastAPI, Next.js, and whatever gets the prototype over the line fastest.
I love an excuse to get coffee.
- Website: leocamacho.co
- Writing: leocamacho.co/essays
- LinkedIn: leo-camacho