AI Engineer

About ETS

ETS is a global education and talent solutions organization enabling lifelong learners worldwide to be future‑ready. For more than 75 years, we've been advancing the science of measurement to build benchmarks for fair and valid skill assessment across cultures and borders. Our worldwide impact extends through our renowned assessments including TOEFL®, TOEIC®, GRE® and Praxis® tests, serving millions of learners in more than 200 countries and territories.

Through strategic acquisitions, we've expanded our global capabilities: PSI strengthens our workforce assessment solutions, while Edusoft, Kira Talent, Pipplet, Vericant, and Wheebox enhance our educational technology and assessment platforms across critical markets worldwide.

Through ETS Research Institute and ETS Solutions, we're partnering with educational institutions, governments, and organizations globally to promote skill proficiency, empower upward mobility, and unlock opportunities for everyone, everywhere. With offices and partners across Asia, Europe, the Middle East, Africa, and the Americas, we deliver nearly 50 million tests annually. Join us in our journey of measuring progress to power human progress worldwide.

The AI Model Development Engineer (open‑rank) supports the TOEFL and GRE assessment programs by designing, developing, evaluating, and deploying machine learning models that power ETS’s next generation of assessment technologies. This role focuses significantly on building and scaling AI‑driven scoring systems for constructed responses, including essays, spoken responses, short answers, and simulations.

Operating at the intersection of AI engineering, assessment science, and operational delivery, the role ensures models are accurate, fair, explainable, and production‑ready. The position contributes to advancing ETS’s legacy in automated scoring, measurement, and assessment innovation. The ideal candidate brings strong applied machine learning expertise, along with experience in model evaluation, data pipelines, and quality controls within high‑stakes or regulated environments.

• Develop, train, and optimize machine learning and deep learning models for applications such as automated scoring (including text, speech, or multimodal responses), item generation, content classification, anomaly detection, and personalization.
• Implement feature engineering, representation learning, and model architectures appropriate for scoring and classification tasks.
• Develop hybrid scoring approaches combining AI models with rules‑based or human‑in‑the‑loop workflows.
• Use NLP, large language models, and multimodal modeling techniques to support assessment creation, delivery, and feedback.
• Build model pipelines, evaluation frameworks, and testing approaches to ensure models are valid, reliable, fair, and aligned with ETS’s Responsible AI guidelines.
• Collaborate with psychometric and validity teams to ensure AI‑driven systems meet technical, fairness, and measurement standards.
• Partner with engineering teams to deploy models into scalable, production‑grade systems integrated with ETS operational platforms.
• Conduct experiments to compare model architectures, datasets, training regimes, and performance tradeoffs.
• Implement monitoring solutions to track model drift, robustness, and security risks over time.
• Participate in cross‑functional design sessions, helping translate assessment or business needs into implementable AI solutions.
• Document model design, training data specifications, evaluation metrics, and deployment requirements.
• Stay current with advancements in LLMs, generative AI, responsible AI, and educational technology, bringing forward ideas for innovation.

• Bachelor’s or Master’s degree in Computer Science, Machine Learning, Data Science, Engineering, or related field.
• 3+ years of experience developing machine learning models in production environments, preferably in large scale scoring environments.
• Hands‑on experience developing and deploying machine learning or deep learning models (Tensor Flow, PyTorch, JAX, or equivalent).
• Experience with NLP techniques and models, including…