Building a collaborative whiteboard classroom from 0 to 20k users — improving student learning at AoPS
role
problem overview
Instructors relied on Zoom for virtual classes and Drawboard for in-person classes, but this experience lacked the collaboration & interactivity central to AoPS learning. Plus, a classroom platform has been the one missing piece in AoPS's end-to-end curriculum delivery.
solution
Building Torchboard, a web-based whiteboard classroom that brings students closer to AoPS's world-class learning material.
An infinite real-time collaborative canvas, so teachers and students can riff on a math problem without ever running out of space.
Interactives like polls with real-time monitoring so students stay engaged and focused on the content at hand
impact
Shipped & supporting over 20,000 students daily, with a 4.3 overall lesson rating.
With less than 1% of users going back to old tools, Torchboard has seen huge success in how teachers teach and how students learn across math, language arts, and science curricula. My vision for Torchboard is to be the default teaching tool for every classroom ever.
context
Art of Problem Solving (AoPS) helps K–12 students develop advanced math, science, language, and computer science problem-solving skills.
AoPS is broadly divided into 3 revenue streams: self-paced classes, learning resources like books and videos, and live classes that are either in-person (at their own campus), virtual (over Zoom), or in a chat-only format. We make our own curriculum and hire our own teachers.
We want to control the entire curriculum delivery lifecycle, especially how the lessons are taught. We expect to completely overhaul how we teach in the next 5 years.
VP R&D, Aops
digging deeper
I observed 25+ live classes & interviewed 15 teachers to understand the learning dynamic.
Teachers mostly relied on Zoom & Drawboard (a PDF annotator) for running their classes, but these tools lacked the collaboration and interactivity central to the AoPS way of learning.
why build our own whiteboard?
As a senior designer on the project, I repeatedly asked this question. While good whiteboards exist (Miro, Mural, Figjam, Freeform, etc.) and they have their own advantages, none of them truly fit for how AoPS teaches. This is why we needed something bespoke.
How might we create a more focused, collaborative & interactive classroom experience for both instructors & students?
Approach
What does the future of AoPS learning look like?
Using my research and aligning with my PM and Eng folks, I carved out design principles that would guide our decision-making.
product roadmapping
While Torchboard started as a replacement for Zoom + Drawboard, we saw a lot of potential in our IRL experiments. So, we leaned into the in-person classroom experience to test out curriculum features and inspire the virtual class components later on.
What if we could do more than just draw?
To truly harness the power of AoPS content, we wanted to create interactives: polls, embeddable PDF readings, images, web links, etc. This would increase student engagement, while also helping teachers rely less on 17 different tabs for each of those things.
early exploration of polls
shipped iteration
What if breakout rooms could be spatial? Like students sitting together for group work?
Teachers struggled to manage, monitor, and support breakout room collaboration at scale. I approached this as a spatial problem where the teacher could see every group's progress in real time and join the group that needs the most help.
early sketches (each mini-board is a breakout)
a lot of experimentation
design systems & design engineering
I built our Design System on top of Shadcn/UI & Tailwind CSS; merged about 50 PRs into our codebase.
There was no org-wide Design System, and our front-end code was inconsistent. So I pushed for Torchboard's Design System to be built on top of Shadcn/ui because it was robust and easy to maintain. Through Cursor and Claude Code, I helped clean up our front-end code.
reflections
Evangelist-Obsessed
Torchboard is a large product and couldn't be launched all at once without risk. Finding great teachers to use it and honing in on their feedback proved to be the best piecemealed launch strategy.
Really impressed with how well Hardik understood the technical side of web dev, and he even shipped some of his own code!
engineering lead, in-class team
Other Featured Work
