nj
← Team
David Zeng
Beacons teamFoundersSan Francisco

David Zeng

Co-founder and Chief Technology Officer

Builds everything a Beacons user can see or touch.

David co-founded Beacons in 2019 with Neal, Jesse, and Greg, after a PhD at Stanford in Dwight Nishimura’s MRI lab, where he built machine learning and signal processing methods to make medical scans faster. His deep learning fix for blurry MRI angiograms became a first-author paper and a Stanford patent. As CTO, his job is, in his words, to build and maintain everything a Beacons user can see or touch. His throughline: creators don’t want better tools, they want fewer problems, and that gap is exactly where AI can win.

At Beacons

  • Builds and maintains everything a Beacons user can see or touch.
  • Building Beacons' AI teammate that vets brand deals and drafts emails for creators.
  • Helped rework how the whole company operates to be AI-native.

Ask David about

  • Why creators want fewer problems, not better tools
  • Rebuilding a company to be AI-native
  • The AI tools he actually uses every week

Fun facts

Background

PhD

Stanford University, Electrical EngineeringDwight Nishimura's Magnetic Resonance Systems Research Lab, 2014 to 2020. Accelerating 3D MR angiography: signal processing for designing RF pulses, and deep learning that corrects off-resonance artifacts so scans can be shorter.

BS

University of Illinois Urbana-Champaign, Electrical Engineering2010 to 2014. Senior thesis with Prof. Zhi-Pei Liang: MRI registration and segmentation for machine-learning diagnosis of Parkinson's disease.

Selected work

Deep residual network for off-resonance artifact correction with application to pediatric body MRA with 3D conesMagnetic Resonance in Medicine, 2019. First author, with Joseph Cheng, Shreyas Vasanawala, Dwight Nishimura, and others. Off-ResNet cut total scan time by 59.3% at image quality non-inferior to standard scans.
Combined T2-preparation and multidimensional outer volume suppression for coronary artery imaging with 3D cones trajectoriesMagnetic Resonance in Medicine, 2020. First author, with Dwight Nishimura and others.
Deep learning method for nonstationary image artifact correctionUS patent 11,681,001, Stanford. Co-inventor with Joseph Cheng, Dwight Nishimura, and Shreyas Vasanawala.
Deep learning method for non-Cartesian off-resonance artifact correctionISMRM-ESMRMB Joint Annual Meeting, 2018. The conference version of Off-ResNet (preprint on arXiv, 1810.00072).
Creators don't want better toolsAI Central Voices interview, 2026.