nothing technology
I build AI and ML powered products at Nothing, end to end. Work at this scale is never one person's: the visual design belongs to Nothing's designers, and everything below shipped with a team around it.
My side runs wider than engineering. I prototype, design systems, and work as an R&D hand, folding an earlier life in design into making the products themselves better; then comes the build, the AI systems, the animation programming, and the glue that turns a design into a product in your hand.
Each section says where my hands were. For anything beyond the public surface, reach out.
Released with Phone (3) in 2025. Essential Search lives one swipe from the home screen: ask it anything and an LLM answers, or it takes you straight to what already lives on your phone, contacts, settings, files. I worked deeply across it, backend and frontend both.
One detail I hold dear. If you have ever tapped to do an essential search, I made that button come alive. The motion design and the icon are the designers' work; mine sits underneath: Android has no built-in system for complex morphing animation, so I engineered one for the button to behave as designed. It ships on Google Play.
2025 to 2026. Most of my time lives in prototypes I cannot show or name. Some of them you may eventually meet in the wild as products; the rest will have quietly made those products better.
- Essential Apps are small, single-purpose apps and widgets that anyone can conjure with a prompt and pin straight to their phone. A gallery collects the community's creations to browse and install. The conjuring is done by a code generation agent I worked on, held in check by novel evaluation systems I created. Evaluation is currently the bottleneck between passable agents and great ones.[1][2] I also worked on the Android app itself, which ships on Google Play.
- Essential Memory is a second memory for your phone: capture what you see, hear and half-think, screenshots, voice notes, moments, and it quietly organises them into something you can come back to and ask questions of. I programmed the app and its UI, building on the team's design.
- [1] Anthropic. Challenges in evaluating AI systems. anthropic.com/research/evaluating-ai-systems
- [2] Yehudai, A. et al. (2025). Survey on evaluation of LLM-based agents. arXiv:2503.16416. arxiv.org/abs/2503.16416