Senior Mobile Engineer – iOS/Android, Realtime AI

We’re hiring a Senior Engineer to build production mobile apps and services that integrate with Meta’s Glasses SDK for RayBan Meta smart glasses. You’ll design and ship hands-free, voice-first, vision-enabled experiences that span the glasses, a companion iOS/Android app, and cloud AI. This is a highly cross-functional role at the intersection of mobile, real-time audio/video, Bluetooth/WiFi transport, and multimodal AI.

• Own end-to-end development of a companion mobile app that interfaces with RayBan Meta smart glasses via Meta’s Glasses SDK

• Implement reliable capture and streaming pipelines for camera preview frames, stills, and multichannel audio, with strict attention to latency, battery, and privacy indicators (camera LED, permissions)

• Build voice-first UX: wakeword handoff, push-to-talk flows, VAD/ASR/TTS, earcons, and low-latency audio playback on openear speakers

• Integrate on-phone computer vision and speech models (Core ML, MediaPipe, ONNX Runtime, NNAPI) and orchestrate cloud inference for multimodal LLMs (e.g., Llama 3family vision/voice) via streaming APIs

• Handle transport and connectivity: Bluetooth LE control channels, WiFi/WiFi Direct media streaming, reconnect logic, and state machines for device pairing and session lifecycles

• Design resilient, observable pipelines with backpressure, retries, offline fallbacks, and graceful degradation when thermals, bandwidth, or permissions change

• Collaborate with product/design on voice-first interaction patterns; run user tests; instrument metrics for latency, accuracy, and task completion

• Establish mobile CI/CD, automated testing (unit/integration/BT device in the loop), crash/error analytics, and release processes

• Champion privacy by design and compliance with platform and bystander safety policies

• 5+ years professional mobile engineering experience, shipping native apps at scale on iOS or Android (preferably both)

• Deep proficiency in:
  • iOS: Swift, SwiftUI/UIKit, AVFoundation, CoreBluetooth, CoreMotion, background modes, concurrency (GCD/AsyncAwait), audio units
  • Android: Kotlin, Jetpack Compose/Views, Bluetooth/BLE, Camera/Media, Foreground services, Coroutines/Flow
• Handson with realtime media pipelines and streaming: audio capture/playback, echo cancellation, noise suppression, lipsync/latency budgeting, WebSockets/gRPC/WebRTC, codecs (Opus/AAC/PCM)

• Solid networking and systems skills: state machines, threading, buffering, backpressure, power/thermal profiling, and debugging on constrained devices

• Experience integrating cloud AI services (LLM/ASR/TTS) and handling streaming inference results in the UI

• Strong product sense for voice-first UX and accessibility in eyes-up, hands-free contexts

• Excellent communication; comfortable working with early/preview SDKs and ambiguous requirements

Preferred qualifications

• Prior work with Meta’s Glasses SDK (RayBan Meta) or similar wearables (Apple Watch/visionOS audio, Snap Spectacles, Bose Frames) – if you don’t have this experience, it is fine, provided that you are willing to learn the Meta Glasses SDK quickly

• Ondevice ML on phone: Core ML/Metal/Accelerate (iOS), MediaPipe/TFLite/NNAPI/GPU delegates (Android), ONNX Runtime Mobile

• Multimodal AI integration: experience with Llamafamily models, Whisper/Seamless/other ASR, TTS providers, prompt and latency optimization, partial results/streaming UX

• BLE expertise: GATT design, connection strategies, MTU/throughput tuning, coexistence with WiFi transport, device provisioning, and firmware update flows

• WebRTC for low-latency A/V; adaptive bitrate, jitter buffers, AEC/VAD tuning

• Backend exposure sufficient to move fast with AI

• Python or Node.js for inference gateways (FastAPI/Express), WebSocket servers, request fanout, and token streaming

• Deploying model servers (Triton, vLLM) and vector/RAG stacks (FAISS/Pinecone), observability (OpenTelemetry), and autoscaling on AWS/GCP/Azure

• Security and privacy: keychain/keystore, secure BLE pairing, PII handling, consent UX, regional data controls

• QA for hardwareintheloop: writing automated tests that exercise glasses events (connect, capture, LED state, battery, IMU), and performance tests for end-to-end latency

Originally posted on Himalayas