AIML - Sr Data Scientist, Data and ML Innovation

We are a group of data scientists, partnering with ML researchers and engineers who develop foundation models that power Apple Intelligence features. We are pushing the boundaries of data science by developing novel techniques to evaluate the performance and capabilities of Foundational Models. In addition, we leverage data mining expertise to identify characteristics of training data that influence the performance of these models. If you are an accomplished data scientist, who wants to expand your influence in this fast evolving and exciting space of Generative AI, this is a great opportunity for you. As a Sr Data Scientist partnering with ML Researchers, you will bring your inquisitive mind and outstanding technical skills to unlock insights about the drivers that improve effectiveness of foundation models. Some projects that you are likely to drive are: - Assess existing foundation models evaluation techniques and improve them to suit Apple’s use cases. - Studying loss patterns of models and attributing them to the drivers across the life cycle of model development from pre-training to SFT to RL. - Defining training data quality metrics and assessing quality of training data to ensure it positively impacts model performance. - Driving interpretability of ablation studies, documenting takeaways and driving a hypothesis driven experimentation strategy. - Building tools and automation process using LLMs to scale data science projects. Minimum Qualifications: 5 years of data science experience demonstrating strong impact to product or model performance. Experience developing evaluation sets and metrics for foundational model performance measurement and diagnostics. Proficiency with applying quantitative methods to structured & unstructured data for exploratory data analysis, pattern recognition, insights generation, metrics development, and scaling analytical tools. Strong programming skills in large scale data manipulation & processing including SQL, Python & Spark. Experience leveraging LLMs in data science workflows. Master’s degree in a technical or quantitative field such as Statistics, Mathematics, Computer Science, Engineering, Economics or Physics. Preferred Qualifications: Deep understanding of foundation model training and development life cycle. Experience with developing model evaluation simulation environments Experience generating synthetic data for foundation model training. Experience with prompt optimization for improving in context learning performance of LLMs. Building and deploying end to end Data Science/ML pipelines. PhD degree preferred.