Machine Learning Engineer - Agentic AI Evaluation Frameworks

Apple•Cupertino, CA

About The Position

The Channel Sales AI Product Engineering team is looking for a Machine Learning Evaluation Engineer to help build and scale evaluation capabilities for our next generation of AI-powered experiences. In this role, you will develop evaluation frameworks, datasets, tooling, and quality signals that enable teams to understand and continuously improve Generative AI and LLM-powered products. You will work closely with Machine Learning, Software Engineering, Quality Engineering, Product, Human Interface, Data Science, and domain experts to establish rigorous evaluation practices throughout the AI product lifecycle. You will help define how we measure the quality of AI experiences across the Commerce domain, including Store AI, Shopping AI, Learning AI, Content GenAI, Conversational AI, and Platform Self-Service. This is an opportunity to work at the intersection of machine learning, software engineering, data, and product quality, helping ensure our AI experiences are accurate, relevant, grounded, reliable, and useful for users around the world.

Requirements

Typically requires a minimum of 7 years of related experience in Machine Learning Engineering, ML Evaluation, Software Engineering, Data Science, Quality Engineering, or a related technical field.
Strong programming skills in Python and experience developing production-quality software, ML systems, data pipelines, or evaluation infrastructure.
Experience developing or evaluating LLMs, Generative AI, Conversational AI, NLP, recommendation systems, or other machine-learning-driven products.
Experience designing automated ML evaluation frameworks, metrics, benchmarks, datasets, or experimentation methodologies.
Understanding of modern LLM application architectures, including prompting, embeddings, retrieval-augmented generation (RAG), tool use, and agentic workflows.
Experience with model-based evaluation techniques and an understanding of the strengths and limitations of approaches such as LLM-as-a-Judge.
Experience with Human-in-the-Loop evaluation, annotation, or data-quality workflows.
Strong understanding of statistical analysis, experimentation, sampling, and measurement methodologies.
Experience performing model error analysis, failure analysis, and root-cause investigation.
Ability to work effectively across Machine Learning, Engineering, Product, Quality, and Data teams.
Excellent written and verbal communication skills, with the ability to translate complex technical findings into clear, actionable recommendations.
Bachelor's degree in Computer Science, Machine Learning, Artificial Intelligence, Data Science, Statistics, Electrical Engineering, or a related technical field, or equivalent industry experience.

Nice To Haves

Experience building evaluation infrastructure for production-scale LLM or Generative AI applications.
Experience evaluating RAG, conversational systems, AI agents, personalization, recommendations, or multimodal AI.
Experience building golden datasets, regression suites, automated quality gates, and continuous evaluation pipelines.
Experience integrating ML evaluation into CI/CD and production release processes.
Experience with prompt evaluation, model comparison, experiment tracking, and AI observability.
Experience evaluating multilingual AI experiences across languages, locales, and markets.
Familiarity with responsible AI evaluation, including robustness, safety, bias, and adversarial testing.
Experience developing internal ML platforms, developer tooling, or self-service evaluation capabilities used across multiple teams.
Experience working with large-scale datasets and distributed ML or data-processing infrastructure.
Master's degree in Computer Science, Machine Learning, Artificial Intelligence, Data Science, Statistics, Electrical Engineering, or a related technical field, or equivalent industry experience.

Responsibilities

Design and build scalable evaluation systems for LLM, Generative AI, Conversational AI, and Agentic AI products.
Design and develop automated evaluation frameworks and pipelines for AI-powered products.
Define evaluation methodologies and quality metrics across dimensions such as accuracy, relevance, groundedness, completeness, consistency, instruction following, and task completion.
Build and maintain high-quality evaluation datasets, including golden datasets, benchmark sets, regression suites, adversarial scenarios, and production-derived test sets.
Develop Auto Eval capabilities that enable teams to rapidly evaluate models, prompts, retrieval systems, agents, and end-to-end AI experiences.
Design and implement model-based evaluation approaches, including LLM-as-a-Judge, while developing appropriate calibration and validation methodologies.
Develop Human-in-the-Loop (HITL) evaluation approaches for complex or subjective quality dimensions where automated evaluation alone is insufficient.
Define evaluation rubrics, annotation guidelines, grading criteria, and quality standards in partnership with product teams, domain experts, and annotation teams.
Build mechanisms to calibrate automated evaluators against human judgment and measure evaluator consistency and reliability.
Evaluate end-to-end AI systems, including retrieval, context construction, prompts, model responses, tool use, APIs, and downstream product experiences.
Develop evaluation methodologies for multi-turn conversations, personalization, recommendations, tool use, reasoning, and agentic task execution.
Perform detailed error analysis and failure-mode investigation to identify opportunities for model, prompt, retrieval, dataset, and product improvements.
Build reusable evaluation infrastructure, APIs, dashboards, and developer tooling that can scale across multiple AI products and teams.
Integrate evaluation into development and CI/CD workflows, enabling automated regression detection, quality gates, and release-readiness assessments.
Connect offline evaluation results with production signals to continuously improve evaluation coverage and product quality.
Partner closely with Machine Learning, Software Engineering, Product, Quality Engineering, Human Interface, and Data Science teams throughout research, development, evaluation, launch, and continuous improvement.