AIML - Machine Learning Engineer - Computer Vision & Audio, MIND

About the Role

The Machine Intelligence, Neural Design (MIND) team, part of Apple’s AIML organization, is leading Apple-wide innovation on HW/SW co-design for efficient inference. With roots in ML, computer vision, and energy efficiency research, our team is strategically positioned to contribute to diverse initiatives ranging from shipping features in well-known Apple products to ambitious, long-term research projects.

We are seeking a hands-on Machine Learning Engineer to drive the data \u0026 evaluation lifecycle for our production models. In this role, you will focus on designing and scaling high-performance data processing pipelines, ensuring data quality, performing in-depth failure analysis on production models, and implementing advanced data augmentation techniques to boost model performance. This includes but is not limited to crafting creative techniques to analyze audio \u0026 video datasets, designing metrics to understand user behavior \u0026 evaluate performance of machine learning models. You will innovate across the entire end-to-end ML production pipeline, bridging the gap between hardware, software, and modeling, ensuring our ML systems are robust, efficient, and scalable.

Description

We are seeking a Machine Learning Engineer to design and deliver innovative features and models that advance our ML systems. In this role, you will scale model evaluation workflows, build robust data pipelines, and optimize performance across the stack.

Your responsibilities will include:

* Pipeline Scaling \u0026 Optimization: Design, build, and maintain scalable ETL/ELT data pipelines using tools like Spark, \u0026 Airflow to handle large-scale datasets. Optimize existing pipelines for efficiency, latency, and cost.

* Data Augmentation \u0026 Synthesis: Research and implement advanced data augmentation techniques (e.g., GANs, semantic augmentation, synthetic data generation) to address data scarcity and imbalanced datasets.

* Data Quality \u0026 Monitoring: Implement data observability and automated data validation checks to identify data drift, schema violations, and outliers in real-time.

* Failure Analysis \u0026 Debugging: Perform root-cause analysis on production model failures, diagnosing issues between data inputs and model outputs using advanced statistical methods.

* Model Evaluation: Collaborate with other machine learning engineers to productize models, implementing robust evaluation frameworks, including experimentation and performance monitoring.

Minimum Qualifications

• Proficiency in working with unstructured data, specifically video \u0026 audio signals, for object detection, pattern recognition, feature extraction and segmentation.
• Proficiency with Python and deep learning frameworks like PyTorch.
• Expertise in designing metrics, and conducting metric change \u0026 performance analysis for model evaluation.
• Strong problem solving skills in analyzing complex, ambiguous problems and clearly presenting sophisticated technical concepts to both expert and non-expert audiences.
• Master’s degree or equivalent experience in a technical or quantitative field.

Preferred Qualifications

• Experience with shipping ML features and products
• Strong verbal and written communications skills with demonstrated experience in authoring \u0026 presenting analytical insights via papers \u0026 presentations.
• Self-motivated and curious with creative and critical thinking capabilities and drive to figure out and improve how things work.
• High tolerance for ambiguity. You find a way through. You anticipate. You connect and synthesize.
• Experience with large scale training ML models including deep learning based models.
• Experience with GPU-based distributed training \u0026 evaluation.
• Background in Computer Vision (image augmentation), Audio and Natural Language Processing.