PhD students Shiyu Wang and Mariam Avagyan have been named recipients of the 2025 Qualcomm Innovation Fellowship (QIF) for North America. The prestigious fellowship supports top graduate students across the U.S. and Canada who are pursuing innovative research with the potential to drive future technologies.
Wang and Avagyan, both PhD candidates in the Electrical Engineering Department and affiliated with the Data Science Institute at Columbia Engineering, earned the award for their research project titled “Fast Manifold Denoising by Learning Traversals.” The research explores efficient denoising techniques on low-dimensional manifolds, with potential applications in areas such as diffusion models and compressed sensing.
“We are deeply honored to receive the Qualcomm Innovation Fellowship,” said Wang and Avagyan in a joint statement. “This support will enable us to further advance our proposed efficient manifold denoising method, which can serve as a building block to larger learning architectures—particularly in scenarios where test-time efficiency is a bottleneck.”
They added: “We are immensely grateful to our advisor, Professor John Wright, for his invaluable guidance, and to Professors Szabolcs Márka and Zsuzsa Márka for their engaging and fruitful discussions. We look forward to connecting with and learning from researchers at Qualcomm.”
Wang and Avagyan’s research focuses on high-dimensional data analysis, signal processing, and machine learning. They aim to design transparent and provable learning methods that utilize hidden data structures to improve computational efficiency. Their approach emphasizes collaboration, first-principle thinking, and developing broadly applicable techniques.
The Qualcomm Innovation Fellowship supports innovative research across key areas such as AI, machine learning, and wireless systems, promoting Qualcomm’s core values of innovation, execution, and teamwork. Learn more about the 2025 North America program here.
