Sandra Delgado, PhD

Learning scientist and builder.

I work at the intersection of pedagogy, AI, and product — turning research into learning experiences that actually work.

About

I’m a learning scientist working at the intersection of AI systems, pedagogy, and product. I help teams build AI-enabled learning experiences where the model behavior, the instructional design, and the product decisions hold together — at scale, and over time.

At Coursera, I’ve shaped the pedagogy and AI behavior behind products used by millions of educators and learners worldwide, including Coursera Coach, Dialogues, Role Play, interactive practice tasks for retrival and recall, and AI Viva exams for academic integrity. Before that, I led curriculum innovation at Universidad Nacional de Colombia, and earned my PhD in Curriculum and Pedagogy from the University of British Columbia, where my dissertation received the Ted T. Aoki Prize for Outstanding Dissertation.

I’m currently pursuing a Master of Engineering in Artificial Intelligence at Universidad de los Andes — because I believe the people designing learning systems should deeply understand the models powering them.

· Working globally · English and Spanish

What I work on

I work on AI-enabled learning systems at the point where cognitive sciences research, product decisions, and long-term AI behavior intersect.

  • AI-mediated learning systems — Practice, feedback, and assessment powered by language models, designed to produce real learning outcomes
  • Prompt and behavior design — Translating pedagogical intent into production-level AI behavior, prompt architecture, and guardrails
  • Learning quality and evaluation — Designing and running evaluations using learner data, outcome metrics, and feedback to measure whether AI-driven learning is actually working
  • Cross-functional product decisions — Working with engineering, design, and product in ambiguous contexts where early AI decisions have long-term consequences

How I work

I’m most effective where teams need to move fast without losing sight of quality. I bring research, data insights, and user feedback into the loop — so decisions about AI behavior are grounded in evidence, not assumptions.

In practice, this means:

  • I make tradeoffs explicit early, instead of letting risk accumulate silently
  • I translate learning science, pedagogy, and model behavior into decision criteria teams can actually use
  • I work cross-functionally to align intent, implementation, and evaluation as systems evolve
  • I focus on long-term system behavior and maintenance
  • I prioritize clarity before scale, knowing it’s the highest-leverage intervention

Selected work

Coursera Role Play

Led the pedagogy and learning design for Coursera’s AI-powered role play feature — enabling learners to practice real-world conversations (difficult feedback, negotiations, complex explanations) with adaptive AI partners. Defined the pedagogical framework, prompt behavior, and evaluation criteria from research through launch. Used learner outcome data and feedback loops to iterate on prompt behavior post-launch.

Coursera Coach & Dialogues

Shaped the pedagogy behind Coursera’s AI learning coach and its dialogue-based learning tool. Developed learning-science-based frameworks for prompt design that increased quality and trust in AI-generated learning facilitation across the platform, with measurable improvements in engagement and learning outcomes based on internal research.

ChatGPT Foundations for Teachers

Collaborated with OpenAI’s education team to design a course helping K–12 teachers build practical AI skills. Worked directly with educators to ground the curriculum in real classroom tasks and workflows — bridging Coursera’s platform with OpenAI’s pedagogical vision.

Global course optimization

Optimized 100+ courses across LATAM and North America using data-driven improvements — integrating generative AI into content workflows, reducing development time by over 50%, and increasing completion rates and learner satisfaction.

Universidad Nacional de Colombia

Led a university-wide pedagogical transformation effort, scaling a learner-centered curriculum model from fewer than 20 to over 100 projects in one year. Co-led a strategic partnership with MathWorks to modernize STEM education through active learning.

How we might collaborate

I collaborate with teams and organizations that are building AI-enabled learning systems and care about cognitive sciences, quality and long-term impact.

  • Strategy and design for AI-powered learning systems
  • Pedagogical and prompt advising for product teams
  • Speaking, panels, or exploratory research on AI in education

Let’s connect

If you’re exploring partnerships, building AI-enabled learning products, or thinking through how pedagogy should shape technology decisions, I’m always happy to exchange ideas.

Email · LinkedIn · Twitter