Sean Bell

CTO

Redwood City, California, United States15 yrs 5 mos experience
Most Likely To SwitchHighly Stable

Key Highlights

  • Director of AI Research at Meta Superintelligence Labs.
  • Co-Founder of GrokStyle, enhancing AI in visual search.
  • Ph.D. in Computer Science with a focus on deep learning.
Stackforce AI infers this person is a leader in AI and Computer Vision within the tech industry.

Contact

Skills

Core Skills

Artificial IntelligenceMachine LearningComputer Vision

Other Skills

LLM pre-trainingdatamix optimizationcurationablationscaling lawssynthetic dataAI for style and designvisual searchrecommendationsinstance recognitionConvNet architecturesmaterial recognitiondeep learningcrowdsourcingmulti-view stereo

About

I am supporting a research org that is all about data for Llama. The single largest lever we have for improving the performance of LLMs is the data that it is trained on. Our group supports all modalities (knowledge, code, reasoning, image, video, speech, ...) and are innovating in areas such as datamix optimization, curation and ablation, scaling laws, understanding the value of data, sourcing, synthetic data. Previously supported GenAI for image generation for ads and monetization. Previously Tech Lead for Product Recognition at Facebook AI. Before then, I was the CEO and Co-Founder of GrokStyle, building AI for style and design. I received my Ph.D. from Cornell in 2016, working on large-scale crowdsourcing, material and object recognition, convolutional neural networks, and deep learning.

Experience

15 yrs 5 mos
Total Experience
3 yrs 3 mos
Average Tenure
7 yrs 4 mos
Current Experience

Meta

Director, AI Research - MSL Data (Meta Superintelligence Labs)

Jan 2019Present · 7 yrs 4 mos · Menlo Park · On-site

  • LLM pre-training (knowledge, code, reasoning, image, video, speech, ...). Datamix optimization, curation and ablation, scaling laws, understanding the value of data, sourcing, synthetic data, and more.
  • Previously: Monetization GenAI
  • Previously: Product Recognition
LLM pre-trainingdatamix optimizationcurationablationscaling lawssynthetic data+2

Grokstyle (acquired by facebook)

Co-Founder and CEO

Sep 2015Jan 2019 · 3 yrs 4 mos · San Francisco, CA

  • AI for style and design. Improving customer experience with AI -- visual search, inspiration, and recommendations.
  • GrokStyle is developing software for visual search to enable instance recognition of an object. The techniques we are developing can be applied broadly to domains like interior design, apparel search, real estate search, product lookup, etc. The company’s current focus is on interior product design. GrokStyle will provide new capabilities to consumers and retailers to assist in matching a consumer’s desired product with the correct or closest matching product based a visual similarity.
AI for style and designvisual searchrecommendationsinstance recognitionComputer VisionArtificial Intelligence

Microsoft

Research Intern

May 2015Aug 2015 · 3 mos · Redmond

  • Advanced the state-of-the-art in object detection, using deep learning
  • Improved detection results by +8.5 mAP over baseline Fast R-CNN on VOC2007
  • Worked with MSR researchers Ross Girshick and Larry Zitnick (now at FB AI Research)
object detectiondeep learningComputer VisionMachine Learning

Cornell university

2 roles

Teaching Assistant

Jan 2015May 2015 · 4 mos · Ithaca, New York Area

  • (Part time) CS4670/5670: Introduction to Computer Vision
  • Gave lectures on deep learning, created course projects, graded assignments, ...
deep learningcomputer visionComputer VisionArtificial Intelligence

Graduate Research Assistant (Ph.D.)

Aug 2011Aug 2016 · 5 yrs · Ithaca, New York Area

  • New ConvNet architectures for visual search across multiple domains
  • Algorithms for material recognition and intrinsic images using deep learning and graphical models
  • Extensible open-source crowdsourcing platform (OpenSurfaces)
  • Multi-view stereo pipeline to reconstruct both geometry and material properties
ConvNet architecturesmaterial recognitiondeep learningcrowdsourcingmulti-view stereoComputer Vision+1

University of toronto

Undergraduate Student Researcher

Sep 2010May 2011 · 8 mos · Toronto, Canada Area

  • Automatically detect noun phrases and find inconsistent references between patent claims
  • Interactive patent editor that provides syntax highlighting and highlights errors in real time
patent detectionsyntax highlighting

Hill & schumacher

Assistant Patent Agent

Jan 2007Jan 2011 · 4 yrs · Toronto, Canada Area

  • (Summer intern for 5 summers)
  • Met with inventors, drafted and reviewed provisional and non-provisional patent applications
  • Prepared responses for examiner reports and office actions, performed patent searches for patentability and freedom to operate, assessed infringement and validity
  • Designed a database and UI to track clients, deadlines, and patent metadata
patent applicationdatabase design

Education

Cornell University

Doctor of Philosophy (Ph.D.) — Computer Science

Jan 2011Jan 2016

University of Toronto

Bachelor's Degree — Engineering Science

Jan 2007Jan 2011

The York School

High School — International Baccalaureate (IB) Diploma

Jan 2007Present

Stackforce found 100+ more professionals with Artificial Intelligence & Machine Learning

Explore similar profiles based on matching skills and experience