Location: Osaka, Japan
Remote: Yes
Willing to relocate: Yes — SF / NYC / Seattle / Tokyo / London (with visa sponsorship)
Technologies: Swift, SwiftUI, UIKit, Metal, Core ML, ARKit, RealityKit, Vision, AVFoundation, visionOS, MLX, llama.cpp, coremltools, ONNX, YOLO, SAM, LaMa, Kotlin / TFLite / NNAPI
Résumé / CV: https://github.com/john-rocky · https://www.linkedin.com/in/daisuke-majima-43a637154/
Email: rockyshikoku@gmail.com
I do on-device ML on iOS. 7 years.
I maintain CoreML-Models (1,745), the de-facto iOS Core ML model zoo. 134 OSS repos covering YOLO, SAM, LaMa inpainting, on-device LLM (CoreML/MLX/llama.cpp), semantic segmentation, super-resolution, and ANE optimization. ios-llm-benchmark publishes A14–A18 inference numbers across runtimes.
Recent shipping: solo lead on a 94,000-line production iOS retail-vision app at Pebble Inc. — ARKit + LiDAR depth + YOLOv8 + CLIP + on-device OCR + multilingual voice input + multi-device backend sync. 70+ TestFlight builds in 6 months, sub-200ms recognition latency on iPhone 15 Pro, in active deployment.
Apps on the App Store under my name (apps.apple.com/developer/daisuke-majima): SnapMeasure (LiDAR measurement utility), Models Zoo (on-device AI), super2x (on-device super-resolution), and others — all on-device.
Career: Pebble Inc. (founder, Oct 2025–present), Ultralytics YOLOv5/v8 team (2024–2025, remote), Agencia Inc. Tokyo (ML engineer, 2022–2024), freelance iOS/CV from 2019.
What I'm looking for: Staff or Senior iOS Engineer / Mobile ML Engineer roles where on-device ML actually matters — Apple Silicon / ANE optimization, real-time CV, AR/LiDAR, on-device LLM, camera/imaging. Available immediately. Open to contracts as a bridge while interviewing for FT.
Japanese (native), English (business-fluent). Happy to talk about ANE quantization, depth/AR coordinate fusion, CoreML conversion for diffusion/inpainting, or what breaks at production scale when you ship iOS solo for half a year.
I do on-device ML on iOS. 7 years.
I maintain CoreML-Models (1,745), the de-facto iOS Core ML model zoo. 134 OSS repos covering YOLO, SAM, LaMa inpainting, on-device LLM (CoreML/MLX/llama.cpp), semantic segmentation, super-resolution, and ANE optimization. ios-llm-benchmark publishes A14–A18 inference numbers across runtimes.
Recent shipping: solo lead on a 94,000-line production iOS retail-vision app at Pebble Inc. — ARKit + LiDAR depth + YOLOv8 + CLIP + on-device OCR + multilingual voice input + multi-device backend sync. 70+ TestFlight builds in 6 months, sub-200ms recognition latency on iPhone 15 Pro, in active deployment.
Apps on the App Store under my name (apps.apple.com/developer/daisuke-majima): SnapMeasure (LiDAR measurement utility), Models Zoo (on-device AI), super2x (on-device super-resolution), and others — all on-device.
Career: Pebble Inc. (founder, Oct 2025–present), Ultralytics YOLOv5/v8 team (2024–2025, remote), Agencia Inc. Tokyo (ML engineer, 2022–2024), freelance iOS/CV from 2019.
What I'm looking for: Staff or Senior iOS Engineer / Mobile ML Engineer roles where on-device ML actually matters — Apple Silicon / ANE optimization, real-time CV, AR/LiDAR, on-device LLM, camera/imaging. Available immediately. Open to contracts as a bridge while interviewing for FT.
Japanese (native), English (business-fluent). Happy to talk about ANE quantization, depth/AR coordinate fusion, CoreML conversion for diffusion/inpainting, or what breaks at production scale when you ship iOS solo for half a year.