iOS Engineer

AGI, Inc.•San Francisco, CA

2d•Onsite

About The Position

Make AGI feel like it was always supposed to be on iPhone. Most "AI on iOS" today is a webview wrapper around a chat completion. That isn't the bar. The bar is: a user picks up their iPhone, the agent does the thing they actually wanted, and the experience feels like Apple shipped it themselves. You'll own the iOS surface for our agent — the app, the keyboard, the share-sheet integrations, the on-device runtime, and every place an Apple user expects an AI agent to show up but mostly hasn't. This is for the engineer who has strong opinions about ProMotion, who reads Apple Human Interface Guidelines for fun, and who can tell within thirty seconds of opening an app whether it was built by people who care.

Requirements

Shipped a meaningful improvement to our iOS app — perf, polish, or a new capability — that a user would notice within 30 days
Owned the iOS roadmap with the CTO and shipped a capability that uses an iOS-specific surface (App Intents, Live Activity, Shortcuts, or Apple Intelligence) in a way that makes the agent feel native within 60 days
Created an iOS experience that can be happily demoed to an Apple engineer within 90 days
Shaped a launch and have a strong, opinionated plan for what an AGI iOS app looks like in twelve months that the team has bought into within 90 days
Strong opinions about ProMotion
Reads Apple Human Interface Guidelines for fun
Can tell within thirty seconds of opening an app whether it was built by people who care

Nice To Haves

Ship iOS that survives Apple review and feels first-party
Read an iOS perf trace and find the actual bug
Run LLMs on Apple Neural Engine at consumer scale
Understand what an agent looks like with full system context vs. inside the Apple sandbox
Understand reliable agentic systems from the people who published the canonical papers on it

Responsibilities

Own the iOS app and SDK end-to-end — architecture, performance, and polish — alongside our CTO
Own our on-device agent runtime on Apple silicon, and the Apple framework integrations (Core ML, MLX, App Intents, Live Activities, Shortcuts)
Manage the "boring-but-existential stuff" — App Store review, entitlements, background execution, battery and thermal profiles
Assist in research by helping land models that run fast and small enough on Apple silicon
Assist in product development by prototyping interaction patterns no one has shipped on iOS before
Assist Android engineers by pushing back when something deserves to be deeply native vs. cross-platform