Tullie Murrell

Tullie Murrell

Whatnot, Shaped, Y-Combinator, Meta, Google, Uber, Palantir

Abstract

I build systems that decide what information matters for humans and agents. Currently leading Applied AI Research at Whatnot and working on all the fun problems related to live commerce marketplace ranking. Before that I founded Shaped, a real-time relevance platform for marketplaces and content companies, and led it as CEO until Whatnot acquired it in 2026. Former Applied Research Scientist at Meta (FAIR); I led PyTorchVideo, was an original core contributor to PyTorch Lightning, and served as technical lead of the fastMRI team. Earlier on in my career I shipped AR and computer-vision products at Meta. Publications in CVPR, NeurIPS, MICCAI, and ACM Multimedia; five patent applications in augmented reality.

I grew up in Adelaide, Australia, spent five years in San Francisco, and now live in New York City. I care deeply about how discovery algorithms shape human behavior and AI research. Outside of the latent space, you'll usually find me hanging with my wife and cats, playing chess, and watching the 49ers.

1   Experience

Director of Engineering, Applied AI Research · Whatnot

2026 to Present

Forming Whatnot's Applied AI Research group to focus on integrating the latest breakthroughs in AI to live commerce. (New York, NY)

CEO & Co-Founder · Shaped

2021 to 2026

Built the relevance layer for the AI world: real-time recommendation and search for marketplaces, e-commerce, and content companies. (New York, NY)

  • Acquired by Whatnot to accelerate AI across live commerce, July 2026 · TechCrunch
  • YC W22; raised $10.7M total; Series A led by Madrona · TechCrunch
  • Customers included Polymarket, Vox Media, QVC, and Outdoorsy
  • Scaled to a 30-person team in NYC

Applied Research Scientist, FAIR · Meta

2019 to 2021

Facebook AI Research (FAIR). Led PyTorchVideo and served as technical lead of the fastMRI research team. (Menlo Park, CA)

  • Lead for PyTorchVideo, a deep learning library powering video understanding across Newsfeed ranking and IG Reels · pytorchvideo.org
  • Original core contributor to PyTorch Lightning · GitHub
  • Technical lead for fastMRI; co-authored 8 papers on MRI reconstruction and generative methods · Meta Engineering

Applied Research Scientist, Building 8 · Meta

2018

Computer vision lead on an AR hardware initiative within Building 8. (Menlo Park, CA)

  • Privileged to work in Building 8, Facebook's moonshot hardware lab · CNET

Machine Learning Engineer · Meta

2016 to 2018

Applied Machine Learning (AML) group. Worked on mobile SLAM, tracking, relocalization, and depth prediction. (Menlo Park, CA)

  • Lead for Facebook World Effects; shipped 3D Drawing and Instagram/Facebook World Effects · CNET
  • Supported Mark Zuckerberg's AR camera keynote at F8 2017 · The Verge

Software Engineering Intern · Google, Uber, Palantir

2014 to 2016

Internships at Google (Search Knowledge Graph), Uber (Driver Incentives), and Palantir (Forward Deployed Engineering). (New York City, Bay Area, Australia)

2   Selected Publications

  1. PyTorchVideo: A Deep Learning Library for Video Understanding . ACM Multimedia
  2. Ego4D: Around the World in 3,000 Hours of Egocentric Video . CVPR
  3. End-to-End Variational Networks for Accelerated MRI Reconstruction . MICCAI
  4. MRI Banding Removal via Adversarial Training . NeurIPS
  5. GrappaNet: Combining Parallel Imaging with Deep Learning for Multi-Coil MRI Reconstruction . CVPR
  6. Advancing Machine Learning for MR Image Reconstruction with an Open Competition: Overview of the 2019 fastMRI Challenge . Magnetic Resonance in Medicine · MRM Editor's Pick
  7. Exploring the Acceleration Limits of Deep Learning Variational Network-based Two-dimensional Brain MRI . Radiology: Artificial Intelligence
  8. Properties of 2D MR Image Reconstructions with Deep Neural Networks at High Acceleration Rates . ISMRM
  9. TorchKbNufft: A High-Level, Hardware-Agnostic Non-Uniform Fast Fourier Transform . ISMRM Workshop
  10. fastMRI: An Open Dataset and Benchmarks for Accelerated MRI . arXiv
  11. Using Deep Learning to Accelerate Knee MRI at 3T: an Interchangeability Study . American Journal of Roentgenology

3   Patents

  1. Sharing and Presentation of Content Within Augmented-Reality Environments US 2020/0066046 A1
  2. Suggestion of Content Within Augmented-Reality Environments US 2020/0066044 A1
  3. Multi-Device Mapping and Collaboration in Augmented-Reality Environments US 2020/0066045 A1
  4. Dynamic Graceful Degradation of Augmented-Reality Effects US 2019/0138834 A1
  5. Systems and Methods for Audio-Based Augmented Reality US 2019/0200154 A1

4   Talks & Podcasts

  1. Getting Generative AI out of development , NYC IDEAS, 2025
  2. How Tullie Murrell built Shaped , Harry Dixon, 2025
  3. How AI and machine learning is shaping what you see , Deals & Reveals, 2025
  4. From Research to Real World AI , The Tech Trek, 2025
  5. The genesis of Shaped and hyper-relevant search in the AI era , OSS4AI Gen AI Zoo, 2024
  6. From Collaborative Filtering to LLMs: Building a Recommendation and Search Platform , OARS @ CIKM, 2024
  7. From Facebook AI to Shaped: Tullie Murrell's Journey to Personalized AI , Avory & Co., 2023
  8. How AI drives Personalization , AI Unleashed, 2023
  9. Robust, accelerated MRI acquisition using AI , RSIP Vision, 2019

5   Links