Software Engineer, on Device Machine Learning

Google Google · Big Tech · Taipei, Taiwan

Software Engineer role focused on adapting and serving AI/ML models for on-device execution, creating AI-powered user-facing features, and optimizing model performance on hardware accelerators. Involves model adaptation, serving infrastructure, quality testing, and customization for specific domains.

What you'd actually do

  1. Adapt models (e.g., computational graph, quantization) for accurate and efficient execution on hardware accelerators (GPU/NPU).
  2. Create end-to-end model serving infrastructures for different applications needs and design AI-powered user-facing features and build app surfaces that deliver them.
  3. Orchestrate quality tests and benchmark performance across a wide variety of models including embedding, Large Language Model (LLM), image inference and image generation.
  4. Customize models for optimal performance in specific domains (e.g., LoRA training).
  5. Provide technical underpinning for production launches such as i18n, provenance and Trust and Safety.

Skills

Required

  • C++
  • C
  • Java
  • Python
  • JavaScript
  • ML infrastructure
  • model deployment
  • model evaluation
  • model serving
  • data processing
  • debugging
  • fine tuning
  • agentic AI development

Nice to have

  • Android machine learning infrastructures
  • Android development
  • advanced AI models
  • LLM
  • quantization
  • hardware acceleration
  • operating systems

What the JD emphasized

  • 1 year of experience with ML infrastructure (e.g., model deployment, model evaluation, model serving, data processing, debugging, fine tuning) or agentic AI development.

Other signals

  • on-device machine learning
  • model adaptation for hardware accelerators
  • end-to-end model serving infrastructure
  • AI-powered user-facing features