Introduction (Bio)

I am a software engineer specializing in On-Device AI and Computer Vision. Formerly an Edge AI Engineer at Ultralytics (creators of YOLOv5/v8), I possess deep expertise in optimizing deep learning models for mobile deployment.

While many engineers focus on cloud-based solutions, I focus on the Edge. I help companies run high-performance AI models directly on iOS/Android devices—reducing server costs, ensuring privacy, and achieving real-time inference without network latency.

Core Competencies (What I do)

• ⚡️ Edge AI Optimization
  • Expertise in model quantization, pruning, and conversion (PyTorch to CoreML/TFLite).
  • Maximizing inference speed on Apple Neural Engine (ANE) and GPU.
• 📱 Native iOS AI Implementation
  • Seamless integration of Vision models (YOLO, Segmentation, Depth) into iOS apps using Swift and C++.
  • Developing custom camera pipelines with AVFoundation and Metal Performance Shaders.
• 🧠 On-Device Generative AI
  • Running LLMs (Llama 3, Gemma) and Diffusion models locally on mobile devices.

Selected Works & OSS

• CoreML-Models (1.7k+ Stars)
  • A collection of SOTA models (YOLO, Depth Anything, SAM, etc.) converted and optimized for iOS.
• SemanticImage
  • A high-performance iOS library for real-time semantic segmentation and image filtering.

Contact / Availability

I am available for freelance consulting and contract engineering. If you need to bring your AI models out of the cloud and into the user's hand, let's talk.

📧 [email protected] 🐦 X (Twitter)

🚀 About

Ultralyticsでの開発経験を持つAIエンジニア。 YOLOなどの最新モデルを、iPhone/iPadなどのエッジデバイスで「実用レベルの速度（Real-time）」で動作させる実装が得意です。 物流、建設、製造業の現場DXを支援します。

📱 Demos

• 物流向け：リアルタイム在庫カウント
  • iPhoneをかざすだけでSKU計測。推論速度15ms。
• 建設向け：ARひび割れ計測
  • マーカーレスで長さ・面積を測定。