Senior Machine Learning Manager, Search & Knowledge Platform

Apple Apple · Big Tech · Santa Clara, CA +1 · Machine Learning and AI

Lead the E2E R&D and engineering for Generative AI models focused on summarization capabilities, including on-device and server-side LLMs, groundedness, and safety models. Develop inference frameworks and integrate with Apple's LLM infrastructure to deliver user experiences across various Apple products.

What you'd actually do

  1. lead the strong team of MLE, SWE, and data engineers responsible for delivering efficient and effective Generative AI models to build and improve the summarization capabilities across different data types.
  2. drive E2E R&D and engineering to generate high-quality summaries and experiences for Apple users.
  3. improve the summarization models’ quality for world knowledge-seeking questions and Safari pages to provide accurate answers and highlight web page gists in real-time.
  4. Lead the team to develop SOTA LLM-based generative models, groundedness models, and safety models for accurate, grounded, concise, and safe summaries.
  5. Devise the product vision and strategy and execute the plan to deliver the highest quality end-user experience.

Skills

Required

  • Machine Learning
  • NLP
  • Generative AI models
  • LLM post-training
  • RLHF/RLAIF
  • reward model
  • RL policy optimization
  • hallucination reduction
  • high availability
  • low latency
  • robustness
  • stability
  • product vision
  • business acumen
  • engineering leadership

Nice to have

  • RAG
  • data mining
  • information retrieval

What the JD emphasized

  • 8+ years of experience in leading engineering/applied research/ML experiences in natural language processing, SOTA generative AI models
  • Proven record of consistent delivery of technology/products across the full Machine Learning life cycle
  • Strong background and experience in Machine Learning, NLP, and RAG.
  • Strong engineering and R&D experience in LLM post-training, advanced RL-based methods to improve LLM models’ safety and quality using RLHF/RLAIF, reward model, advanced RL policy optimization algorithms, cutting-edge hallucination reduction methods, and their engineering implementation, hands-on experience to develop and ship RL based models with high availability, low latency, robustness, and stability.

Other signals

  • Generative AI models
  • on-device LLM models
  • summarization capabilities
  • Private Cloud Compute servers
  • SOTA LLM-based generative models
  • groundedness models
  • safety models
  • on-device and on-server software frameworks
  • LLM-based model inference
  • Apple ecosystem with Apple’s LLM infrastructure
  • Apple’s LLM infrastructure and generative models