About the Role
Responsibilities
Partner with ML Engineers to own conversion of python features to on device.
Key Activities
Software Development (Swift implementation, testing, debugging, documentation)
ML-to-Device Conversion (Core ML optimization, Foundation Models integration, performance profiling)
Tasks
Read initiatives, review wireframes, process diagrams, and data requirements in order to draft solution options during the architecture review including level of effort for implementation.
Collaborate with the Platform Architect on those options to ensure they follow best practices.
Translate the options into high code quality that adheres to Swift and Honestum best practices. Write tests to ensure code quality and prevent regressions.
Participate in grooming calls to get clarification on the business requirements and execute sprint demos to share progress. Work with business planning on the talk track of these demos
Debug issues found during user acceptance testing (UAT). Including monitoring and debug performance issues (load times, memory, crash rates, etc). Escalate to Platform Architect when blocked.
Document feature during technical handover to ensure the platform is maintainable.
Convert ML models from Python/PyTorch (>1GB) to Core ML (<300MB) using quantization and pruning techniques
Prototype ML features on Apple hardware with thermal/memory profiling, achieving 0.8ms time-to-first-token and 30 tokens/sec on iPhone 15 Pro
Implement Foundation Models framework with 4,096-token context window management, session resets, and GenerationError handling
Maintain <30% thermal throttling during sustained ML workloads across Apple Intelligence devices (A17 Pro+, M-series, M1+)
Configure PrivacyInfo.xcprivacy manifests and Foundation Models Adapter Entitlements for App Store compliance
Design tool calling patterns as primary extension mechanism over adapter training for production use cases
Skills Set (Must Have)
5+ iOS/macOS development with 2+ years Core ML and on-device ML experience
Proficiency with Apple ML frameworks (Core ML, Create ML, Accelerate) and Instruments for performance profiling demonstrated in a track record of shipped ML features with proven success for thermal, memory, and battery constraints
Experience optimizing ML models through quantization, pruning, and compression (PyTorch/TensorFlow to Core ML)
Experience with LLM concepts: token management, context windows, prompt engineering
Ability to rapidly prototype and evaluate ML approaches against device constraints
Knowledge of Apple Intelligence architecture and privacy requirements
Advanced Swift/SwiftUI skills managing complex async operations and state across multiple Apple platforms
Skills Set (Nice to Have)
Hands-on experience with Foundation Models framework (iOS 18+/macOS 15+)
Experience with Background Assets framework and adapter management