Sr Machine Learning Engineer

Disney Entertainment & Sports LLC, working with ESPN, is seeking a Sr Machine Learning Engineer based in New York, NY (onsite) to lead the end-to-end journey of advanced deep learning models. This role focuses on designing, building, and productionizing models used in large-scale recommendation and ranking systems, with ownership spanning architecture, training methodology, offline evaluation, inference optimization, and online serving code.

The position partners closely with platform and infrastructure teams to ensure successful deployment and reliable operations. Responsibilities also include defining model interfaces and observability signals, coordinating safe release criteria, and contributing to technical standards and design reviews for deep learning and recommender-system development.

What you’ll do

• Lead design and implementation of advanced deep learning models for retrieval, ranking, and personalization, including embedding-based, sequence, transformer, and multi-task architectures.
• Convert product and personalization goals into model formulations, loss functions, sampling strategies, feature representations, and evaluation plans grounded in machine learning theory.
• Build production-grade training and inference code with emphasis on correctness, modularity, testing, reproducibility, and maintainability.
• Optimize online performance for latency, throughput, memory efficiency, and cost, using techniques such as distillation, quantization, pruning, batching, compilation, and efficient architecture design where appropriate.
• Own the productionization path from prototype through offline validation and online integration, ensuring consistency between training and inference behavior.
• Design offline experiments, ablation studies, and evaluation frameworks, partnering with data science and product teams to measure business impact with online experiments.
• Diagnose model behavior and failure modes, including bias, calibration, cold start, feedback loops, delayed labels, distribution shift, and relevance-quality tradeoffs.
• Collaborate with MLOps, platform, data, and serving teams to define model interfaces, runtime requirements, observability signals, and safe release criteria (without owning general ML infrastructure).
• Participate in technical design reviews, help set model engineering standards, and mentor engineers on deep learning, recommender systems, and production-quality development.

Requirements

• Strong theoretical and applied knowledge of deep learning, including optimization, representation learning, regularization, generalization, and modern neural architectures.
• Proven experience designing and launching deep learning models in production, with ownership beyond notebook experimentation to production-quality implementations.
• Hands-on experience with recommendation, ranking, retrieval, search, or personalization models at meaningful scale.
• Expert proficiency in PyTorch, TensorFlow, or a comparable deep learning framework, plus strong Python software engineering skills.
• Experience improving inference performance and efficiency, including practical understanding of latency, throughput, memory, hardware utilization, and model-quality tradeoffs.
• Experience with large-scale model training, distributed data processing, and production model evaluation.
• Solid understanding of experimentation and evaluation, including metric selection, offline-online correlation, ablation analysis, and statistical validation.
• Ability to communicate complex modeling decisions clearly and collaborate effectively with product, data science, data engineering, platform, and infrastructure partners.
• Experience with multi-stage recommender systems, including candidate generation, retrieval, ranking, re-ranking, and embedding-based approaches.
• Deep expertise in one or more areas such as transformers, sequence modeling, multi-task learning, graph neural networks, self-supervised learning, contextual bandits, or reinforcement learning for recommendations.
• Experience optimizing neural models with ONNX, TensorRT, CUDA, custom kernels, or similar inference acceleration technologies.
• Ability to design models for strict real-time serving constraints and measure p95/p99 latency, throughput, memory footprint, and cost-to-serve.
• Knowledge of recommender-system challenges such as implicit feedback, negative sampling, selection bias, cold start, exploration versus exploitation, and delayed outcomes.
• Research publications, patents, open-source contributions, or equivalent evidence of depth in deep learning or recommender systems.
• Experience working in fast-paced or 0-to-1 environments with ambiguous requirements.
• 5+ years of hands-on experience developing machine learning systems, including substantial experience designing and productionizing deep learning models.
• Bachelor’s or Master degree in Computer Science, Machine Learning, Data Science, or a related field, or equivalent practical experience.

Technologies

• PyTorch
• TensorFlow
• Python
• ONNX
• TensorRT
• CUDA

Pay transparency

• The hiring range for this position in New York, NY is $150,900-$202,400, and for Glendale, CA is $144,100-$193,200.
• The base pay actually offered will consider internal equity and may vary depending on geographic region, job-related knowledge, skills, and experience.
• A bonus and/or long-term incentive units may be included, in addition to full-range medical, financial, and/or other benefits, depending on the level and position offered.