An AI Ally for Teachers
At Georgia Tech’s 2025 Career, Research, and Innovation Development Conference, PhD student Shi Ding didn’t expect to walk away with a startup idea. She was presenting a research poster on accessibility-focused AI when a school principal, one of the judges, asked if she had ever considered commercializing her work. The question lingered.
Encouraged by that exchange, Ding applied to Georgia Tech’s CREATE-X startup accelerator and was selected for the Quadrant-I cohort, a program designed to help graduate researchers translate academic work into real-world ventures. While continuing her research, she stepped into Atlanta’s startup ecosystem. “It was the energy that motivated me,” Ding says. “You’re in this vibrant environment where everyone is building something.”
Quadrant-I pushed Ding to identify a concrete problem worth solving. She works out of the Technology Square Research Building as a researcher in Georgia Tech’s Digital Media department under Professor Brian Magerko in the Expressive Machinery Lab. Her background includes education-focused projects such as EarSketch, which combines coding and music to teach computer science, as well as experience as a cognitive scientist intern at an AI startup during her second year of doctoral studies.
Drawing on that mix of research and industry exposure, Ding began interviewing K–12 teachers, teaching assistants, and university instructors. The feedback was consistent: creating instructional materials, especially ones that can be safely and responsibly used in classrooms, consumes an enormous amount of time. “Creating instructional materials that can be safely used in real-world classrooms often takes educators an extensive amount of time,” Ding says. “And general-purpose AI tools like ChatGPT aren’t built for domain-specific stakeholders like educators.”
Building Deeproject AI
Ding founded Deeproject AI to address that gap. The startup is developing a trustworthy AI companion designed specifically for education—one that integrates instructional safety mechanisms and aligns closely with pedagogical practice. Teachers can input topics, grade levels, and learning objectives to generate complete, standards-aligned lesson plans.
“We built Deeproject to be an educator’s co-pilot. It takes the heavy lifting off their plate so they can focus on teaching.”
Ding leads product development and research, while her co-founder, Sijian “Charles” Tan, a computer science engineering graduate student at Georgia Tech, leads engineering. Together, they are building an AI system that supports project-based learning, producing materials that include lesson plans, rubrics, activities, and reflection prompts. “Too often, hours are lost drafting lessons and scaffolding,” Ding says. “Our AI gives those hours back.”
By the end of the CREATE-X accelerator, the team had a working prototype, which they showcased live at Demo Day. Since then, Deeproject has begun testing the tool with a small group of educators. “Right now, our focus is user feedback, technical iteration, and finding the right partners,” Ding says.
Gaining Ground in Tech Square
As a CREATE-X startup, Deeproject has received early support from Microsoft, AWS, and Google through startup credits and cloud resources. “That support was huge for us,” Ding says. “It showed others believed in what we were building.”
Ding has also continued to validate Deeproject’s underlying research alongside her academic work. In 2025, she co-authored and presented research with Magerko at the NeurIPS Foundation Models Workshop, focused on evaluating AI agents and alignment. She has also contributed to an archival publication accepted at SIGCSE, further reinforcing the team’s emphasis on safety, pedagogy, and responsible AI in education.
With ongoing mentorship from the Quadrant-I network, including entrepreneur-in-residence Richard Gruber, the team is actively exploring partnerships with schools, training organizations, and research agencies. Their approach emphasizes co-development with educators rather than top-down deployment. “We want to build this with human needs in mind—not just for them,” Ding says.
What began as a research poster has grown into a working startup with real classroom potential. Deeproject AI reflects the strength of Tech Square’s innovation ecosystem; one where academic ideas do not remain confined to papers or posters, but evolve into tools with tangible educational and societal impact.