This is not a standard Android engineering role. You will be working across two surfaces simultaneously — a standard Android mobile app that mirrors an iOS experience, and an AOSP-based custom Android build running on wearable hardware. You will be making foundational decisions about how an agentic AI platform feels and behaves in a wearable form factor — a problem that has no established conventions and very few reference points.
You should also know that the wearable platform will expand from AOSP to RTOS asap. The right candidate for this role is excited by that transition rather than daunted by it — someone who thinks about platform architecture and is comfortable moving between layers of the stack as the hardware evolves.
You will work directly with the CEO and a small, high-trust engineering team across Silicon Valley, Paris, including close collaboration with a China-based firmware team who own the hardware bring-up.
We actively use AI development tools across our engineering team — Cursor, Claude, Copilot — and expect engineers who use them seriously as a core part of their workflow.
What You'll Build
A standard Android mobile app at feature parity with our existing iOS experience — bringing the Agentic OS to Android users with the same interaction quality and design integrity
An AOSP-based client for wearable hardware — building the application layer on top of a custom Android stack running on a constrained wearable form factor
The agent interaction surface for wearables — defining what it means to interact with an always-on AI when there is no touchscreen, no keyboard, and no conventional UI paradigm to fall back on
A platform architecture that is designed to survive the transition from AOSP to RTOS — thinking ahead so that the work done today does not need to be discarded when the platform evolves
Close integration with the firmware team — bridging the application layer and the hardware bring-up work happening in Shenzhen
On-device processing integrations — working to minimize cloud dependency and maximize what runs locally on constrained hardware