Senior Applied Researcher

Apple Apple · Big Tech · San Francisco, CA +1 · Machine Learning and AI

Senior Applied Researcher with expertise in Generative AI, LLM architectures, and advanced NLP systems. The role involves architecting, designing, and deploying LLM-powered systems for personalization, automation, and customer understanding across Apple Services. Responsibilities include research in representation learning, semantic modeling, NLU, RAG, fine-tuning, evaluation, safety alignment, and exploring methods like parameter-efficient adaptation, multi-agent orchestration, and RLHF. The role also involves building prototypes and production-grade solutions, contributing to patents/publications, and mentoring other researchers. Requires a Ph.D. or equivalent experience, expert knowledge of deep learning/NLP/transformers, LLM development, Python, and ML frameworks, with experience deploying models in production.

What you'd actually do

  1. Architect, design, and deploy LLM-powered systems that unlock new capabilities for personalization, intelligent automation, and customer understanding across Apple’s services ecosystem.
  2. Lead research in areas such as large-scale representation learning, semantic modeling, topic induction, natural language understanding, and retrieval-augmented generation (RAG).
  3. Drive LLM fine-tuning, evaluation, safety alignment, and optimization strategies to ensure performant, compliant, and frictionless user experiences.
  4. Explore and productize methods such as parameter-efficient adaptation, multi-agent orchestration, active learning, RLHF, and novel inference optimization techniques.
  5. Build prototypes and production-grade solutions that advance Apple’s ability to reason over text, behavioral signals, and domain-specific knowledge at scale.

Skills

Required

  • Generative AI
  • LLM architectures
  • advanced NLP systems
  • deep learning
  • modern NLP
  • transformer architectures
  • foundation model adaptation
  • LLM model development
  • fine-tuning
  • instruction tuning
  • prompt engineering
  • Python
  • ML frameworks (PyTorch or TensorFlow)
  • deploying models in production systems
  • distributed data processing systems (e.g., Spark)
  • large-scale experimentation

Nice to have

  • Retrieval-augmented generation (RAG) pipelines
  • vector-based semantic search systems
  • Representation learning
  • semantic embeddings
  • clustering
  • categorization
  • content understanding
  • Model evaluation frameworks
  • language quality
  • relevance
  • hallucination
  • safety
  • Inference optimization techniques
  • quantization
  • distillation
  • model compression
  • reinforcement learning
  • policy alignment
  • RLHF
  • interactive AI systems
  • personalization algorithms
  • ranking algorithms
  • optimization algorithms
  • architectural design
  • RL experiment platform
  • multi-armed bandit experiment platform
  • publications in top-tier ML/AI venues
  • patent filings

What the JD emphasized

  • taken novel ideas from research inception to production deployment
  • influence and implement foundational intelligence capabilities
  • operate with the autonomy and scope expected of senior technical leadership
  • LLM fine-tuning, evaluation, safety alignment, and optimization strategies
  • parameter-efficient adaptation, multi-agent orchestration, active learning, RLHF, and novel inference optimization techniques
  • reason over text, behavioral signals, and domain-specific knowledge at scale
  • Ph.D. in Computer Science, Machine Learning, NLP, Statistics, or a related field—or equivalent industry experience delivering production AI systems
  • Proven ability to communicate research outcomes, architectural decisions, and technical tradeoffs to technical and non-technical stakeholders
  • A record of publications in top-tier ML/AI venues or patent filings demonstrating novel research contributions

Other signals

  • LLM-powered systems
  • large-scale representation learning
  • semantic modeling
  • natural language understanding
  • retrieval-augmented generation (RAG)
  • LLM fine-tuning
  • evaluation
  • safety alignment
  • optimization strategies
  • parameter-efficient adaptation
  • multi-agent orchestration
  • active learning
  • RLHF
  • novel inference optimization techniques
  • reason over text, behavioral signals, and domain-specific knowledge at scale