In this role you will have the opportunity to develop LLMs and other NLP models for user productivity and improving Siri's ability to answer personal domain questions. Our team owns models which are responsible for answering users' questions using their personal documents with privacy at the forefront, and that integrates with other Siri capabilities to enable powerful user experiences. Role responsibilities include:
- Contribute to research, design, implementation, and evaluation of models to enhance quality and performance, and to support key functions for Personal Q&A.
- Implement and extend LLM models with fine-tuning and RLXF methodologies. Develop high quality data pipelines to support these methodologies.
- Establish target quality metrics and evaluation sets for the team to track and improve. Design novel experiments to validate/refute hypotheses and support team wide decision making.
- Collaborate with partner teams to define product requirements and priorities, and to explore opportunities for enhancements to the Personal Q&A stack.
- Develop long-term technical vision for Personal Q&A quality; identify problem areas and integrate solutions as part of a larger roadmap.
8+ years industry experience in Machine Learning, NLP and applying these techniques at scale
Strong software engineering skills in mainstream programming languages, such as: Python, Go, C/C++
Experience using ML frameworks (pyTorch, JAX, TensorFlow, XGBoost etc.)
Strong communication skills
Bachelors in Computer Science
In-depth knowledge and expertise in training, evaluating, and applying Deep learning models, Large Language Models for production systems
Extensive experience in building production ML systems and applications in search, recommendation systems, or information retrieval
Ability to quickly prototype ideas / solutions, and perform critical analysis
Background in: search relevance and ranking, Q&A, personalization, user behavior/response modeling, or data-driven decision-making
Advanced degree (Master’s or Ph.D.) in Computer Science, Statistics, or related field, or equivalent industry work experience