Skip to content
Prompt design and system architecture for the AI matching logicDevelopment of a custom evaluation framework, using LLMs to simulate quiz completions across 350+ degrees to calculate hit rates and surface gapsBackend content alignment: partnered with content and UX teams to optimize degree pages for AI compatibilityLed enrollment coach enablement: created and delivered training, FAQs, and support materials so coaches could confidently talk to students about how DegreeMe worked and what the results meant
OpenAI APIs for natural language generationA RAG (retrieval-augmented generation) pipeline that indexes and vectorizes all degree contentA Cohere re-ranker model to fine-tune top matches
DegreeMe
Helping prospective students find the right ASU degree - fast.
🔍 Overview
DegreeMe is an AI-powered degree-matching tool built to simplify the decision-making process for prospective ASU Online students. With over 350 degree programs available, many students were experiencing choice paralysis and dropping off the site before discovering programs that could be a perfect fit. DegreeMe flips that experience on its head with a quick, quiz-style interface that uses generative AI to surface personalized degree recommendations and explain why those choices make sense.
🧠 My Role
As Product Manager for the AI system, I led:
I worked cross-functionally with designers, content strategists, engineers, and another PM focused on the website experience to bring this from concept to launch.
🎯 The Challenge
UX research revealed a consistent problem: users were browsing ASU’s degree catalog alphabetically, rarely making it past programs starting with "C." Filters were underutilized, and students were overwhelmed by volume — not disinterest. This led to high drop-off rates and lost opportunities for engagement. We needed a way to make discovery feel human, helpful, and fast.
💡 The Solution
DegreeMe replaced passive browsing with a guided, 10-question quiz designed to capture a student’s interests, goals, and preferences. Once submitted, the system uses:
Students receive a shortlist of five degrees, each paired with a personalized description explaining why that program is a strong fit — all generated dynamically using LLMs.
📊 Results (First 90 Days)
8,300+ quiz completions
900 RFIs submitted, with a 10.9% RFI conversion rate (vs. 3.9% site-wide)
400+ submitted applications tied to DegreeMe usage
4.9% application start rate vs. 4.7% site baseline
3.7/5 match satisfaction score and 79/100 usability score from post-surveys
Students reported discovering programs they “never knew existed,” with the experience described as “fun,” “easy to use,” and “super helpful.”
🛠️ Tools Used
Coda – product documentation, user stories, decision logs, project tracking
OpenAI + Cohere – LLM integration and re-ranking
Figma, Slack, Notion, ASU internal tooling – design and comms
🎓 Enrollment Coach Enablement
To support the rollout of DegreeMe, I led a training session for ASU’s Enrollment Center coaches, the frontline staff who guide prospective students through their decision-making process. Since DegreeMe introduces a new AI-powered experience into that journey, it was critical to ensure coaches understood how the tool worked, how recommendations were generated, and how to answer common student questions.
I walked the team through the technology behind DegreeMe, explained its role in surfacing tailored degree options, and provided a detailed rollout overview. The goal was to equip every coach with the context and confidence to integrate DegreeMe into their conversations, reinforcing trust in the tool and enhancing the student experience.
📈 Post-Launch Analysis
Following the launch of DegreeMe, I collaborated with a cross-functional team of product managers, data scientists, and UX researchers to deliver both 30-day and 90-day post-launch analysis reports. These reports highlighted key usage metrics, including total completions, RFI conversion rates, and application outcomes, along with qualitative feedback from students. Each analysis helped surface actionable insights and informed next-step recommendations to further improve tool performance, student experience, and internal alignment.
Up Next:
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.