The RTutor Project
Shipped Product
AI Product Design
UX Research
Product Management
what is rtutor?
High-tech calculator. Glorified translator.
RTutor is an AI-powered tool that turns pseudocode into R code. It is a required part of STAT 100 at UIUC, an intro stats class with 1000+ students a semester.
The issue? Students hated using it.
My job was to find out why and fix it.
High-tech calculator. Glorified translator.
RTutor is an AI-powered tool that turns pseudocode into R code. It is a required part of STAT 100 at UIUC, an intro stats class with 1000+ students a semester.
The issue? Students hated using it.My job was to find out why and fix it.
My Mentor: Dr. V.N. Vimal Rao
High-tech calculator. Glorified translator.
RTutor is an AI-powered tool that turns pseudocode into R code. It is a required part of STAT 100 at UIUC, an intro stats class with 1000+ students a semester.
The issue? Students hated using it.
My job was to find out why and fix it.
raw data -> results
This was the original RTutor website. Users can upload a dataset and write prompts, which generate graphs, analyses, and R code.
I surveyed 726 students in STAT 100.
I pulled from established UX research methodologies, including meCUE, SUS, and UEQ.
Key Findings:
Performance was the main issue for students.
Students wrote a prompt → the translated code crashed → the site crashed.
Students think RTutor is valuable and helps them reach their learning goals.
We realized that students were writing pseudocode incorrectly. The generated R code then crashed, causing the whole site to crash.
My research is ongoing. Check back later for updates!
Preliminary Findings:
➢ Performance issues (crashes, slow runtime) were the main factors for a negative perception
➢ Students are split on whether the website is easy to understand
➢ Despite negative emotional responses, students consistently rank RTutorAI as practical and valuable
Potential Design Improvements:
➢ Improve performance (crashes, runtime) to handle high website traffic
➢ Redesign the "Submit" and "Reset" buttons
➢ Display past prompts, along with output history
➢ Redirect users when they get errors or null responses
➢ Incorporate more instructions or add a tutorial
➢ Allow students to ask questions about the R code or graphs
➢ Add the ability to manually modify code
What's Next:
➢ Analyze and organize the data for my professor's paper
➢ Act as a UI/UX consultant, deciding which changes could be implemented
➢ Design mockups to hand off to developers
➢ Conduct future surveys and UX research to ensure the changes are effective
This project has shown me how many factors go into a product's success. These data-driven changes have the potential to make a real impact on student learning and motivation. This is an incredible responsibility and I can't wait to keep working!
Thank you to V.N. Vimal Rao for overseeing and mentoring this project.
We received over 1000 pieces of write-in feedback.
I organized them into 4 categories (with 22 subcategories):
System Functionality
User-AI (HCI) Interaction
UI - User Input
UI - Interpreting Output
These eventually became the product requirements.
My research is ongoing. Check back later for updates!
Preliminary Findings:
➢ Performance issues (crashes, slow runtime) were the main factors for a negative perception
➢ Students are split on whether the website is easy to understand
➢ Despite negative emotional responses, students consistently rank RTutorAI as practical and valuable
Potential Design Improvements:
➢ Improve performance (crashes, runtime) to handle high website traffic
➢ Redesign the "Submit" and "Reset" buttons
➢ Display past prompts, along with output history
➢ Redirect users when they get errors or null responses
➢ Incorporate more instructions or add a tutorial
➢ Allow students to ask questions about the R code or graphs
➢ Add the ability to manually modify code
What's Next:
➢ Analyze and organize the data for my professor's paper
➢ Act as a UI/UX consultant, deciding which changes could be implemented
➢ Design mockups to hand off to developers
➢ Conduct future surveys and UX research to ensure the changes are effective
This project has shown me how many factors go into a product's success. These data-driven changes have the potential to make a real impact on student learning and motivation. This is an incredible responsibility and I can't wait to keep working!
Thank you to V.N. Vimal Rao for overseeing and mentoring this project.
prompt evaluations:
a student-centered design
We want students to use their critical thinking skills. AI can fix mistakes, but we tell RTutor not to.
Since the biggest problem was crashing, we introduced prompt evaluations.
I designed a feature that checks each prompt to make sure it has enough detail — and if it doesn't, RTutor sends it back for the student to fix.
My research is ongoing. Check back later for updates!
Preliminary Findings:
➢ Performance issues (crashes, slow runtime) were the main factors for a negative perception
➢ Students are split on whether the website is easy to understand
➢ Despite negative emotional responses, students consistently rank RTutorAI as practical and valuable
Potential Design Improvements:
➢ Improve performance (crashes, runtime) to handle high website traffic
➢ Redesign the "Submit" and "Reset" buttons
➢ Display past prompts, along with output history
➢ Redirect users when they get errors or null responses
➢ Incorporate more instructions or add a tutorial
➢ Allow students to ask questions about the R code or graphs
➢ Add the ability to manually modify code
What's Next:
➢ Analyze and organize the data for my professor's paper
➢ Act as a UI/UX consultant, deciding which changes could be implemented
➢ Design mockups to hand off to developers
➢ Conduct future surveys and UX research to ensure the changes are effective
This project has shown me how many factors go into a product's success. These data-driven changes have the potential to make a real impact on student learning and motivation. This is an incredible responsibility and I can't wait to keep working!
Thank you to V.N. Vimal Rao for overseeing and mentoring this project.
design -> development
In total, I wrote out 33 product requirements. Our amazing developer, Jenna, worked with us to implement them all in six months. Here are a few:
My research is ongoing. Check back later for updates!
Preliminary Findings:
➢ Performance issues (crashes, slow runtime) were the main factors for a negative perception
➢ Students are split on whether the website is easy to understand
➢ Despite negative emotional responses, students consistently rank RTutorAI as practical and valuable
Potential Design Improvements:
➢ Improve performance (crashes, runtime) to handle high website traffic
➢ Redesign the "Submit" and "Reset" buttons
➢ Display past prompts, along with output history
➢ Redirect users when they get errors or null responses
➢ Incorporate more instructions or add a tutorial
➢ Allow students to ask questions about the R code or graphs
➢ Add the ability to manually modify code
What's Next:
➢ Analyze and organize the data for my professor's paper
➢ Act as a UI/UX consultant, deciding which changes could be implemented
➢ Design mockups to hand off to developers
➢ Conduct future surveys and UX research to ensure the changes are effective
This project has shown me how many factors go into a product's success. These data-driven changes have the potential to make a real impact on student learning and motivation. This is an incredible responsibility and I can't wait to keep working!
Thank you to V.N. Vimal Rao for overseeing and mentoring this project.
rtutor returns to the classroom
Students will be using the new RTutor starting this fall! I will continue as PM of UIUC RTutor to monitor its performance.
I will also be writing a paper with my mentor, Dr. V.N. Vimal Rao, on RTutor and the design of AI educational tools.
