Research

I am a Principal Scientist at Amazon with 20+ years of experience at the intersection of AI research, generative and multimodal models, computer vision, and computational photography. I transform advanced research into scalable products used by millions of customers worldwide. My work spans learned representations, 3D human understanding, camera geometry, and computational image formation, with 50+ U.S. patents and 55+ peer-reviewed publications in top-tier venues including CVPR, ICCV, ECCV, and SIGGRAPH. With 12,000+ citations and an h-index of 50, my research reflects a sustained record of innovation and impact. I thrive on bridging cutting-edge AI research with customer-focused product delivery, leading teams from concept to deployment.

Research areas

Generative AI & visual learning

Learning to create and understand visual data

Research includes synthetic data generation, weakly supervised learning, semantic segmentation, and learned visual representations.

3D vision & human understanding

Shape, pose, and scene structure

Work on 3D human pose and shape estimation, reconstruction from sparse views, geometric self-supervision, and pose estimation.

Computational imaging

Co-designing cameras and computation

Computational photography, coded exposure, light fields, motion-aware imaging, camera calibration, and image formation.

Selected Papers

FlexNeRF research figure

CVPR 2023FlexNeRF: Photorealistic Free-viewpoint Rendering of Moving Humans from Sparse Views

Multi-Instance Pose Networks research figure

ICCV 2021Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation

PoseNet3D research figure

3DV 2020PoseNet3D: Learning Temporally Consistent 3D Human Pose via Knowledge Distillation

Box2Seg research figure

ECCV 2020Box2Seg: Attention Weighted Loss and Discriminative Feature Learning for Weakly Supervised Segmentation

Pose estimation research figure

CVPR 2019Unsupervised 3D Pose Estimation with Geometric Self-Supervision

Synthetic compositing research figure

CVPR 2019Learning to Generate Synthetic Data Via Compositing

Context encoding research figure

CVPR 2018Context Encoding for Semantic Segmentation

CRAFT research figure

ECCV 2018 workshopCRAFT: Complementary Recommendations Using Adversarial Feature Transformer

Can 3D Pose be Learned from 2D Projections Alone? research figure

Best Paper, ECCV 2018 PeopleCap Workshop Can 3D Pose be Learned from 2D Projections Alone?

Extrinsic Camera Calibration Without A Direct View Using A Spherical Mirror research figure

ICCV 2013 Extrinsic Camera Calibration Without A Direct View Using A Spherical Mirror

Towards Motion-Aware Light Field Video For Dynamic Scenes research figure

ICCV 2013 Towards Motion-Aware Light Field Video For Dynamic Scenes

Single Image Calibration of Multi-Axial Imaging Systems research figure

CVPR 2013 Single Image Calibration of Multi-Axial Imaging Systems

$Checkerboards viewed through a water tank for flat refractive camera calibration$

CVPR 2012 A Theory of Multi-Layer Flat Refractive Geometry

Motion-Aware Structured Light using Spatio-Temporal Decodeable Patterns research figure

ECCV 2012 Motion-Aware Structured Light using Spatio-Temporal Decodeable Patterns

Convex Bricks: A New Primitive for Visual Hull Modeling and Reconstruction research figure

ICRA 2012 Convex Bricks: A New Primitive for Visual Hull Modeling and Reconstruction

Forward projection research figure

ECCV 2010 Analytical Forward Projection for Axial Non-Central Dioptric and Catadioptric Cameras CVPR 2011 Beyond Alhazen’s Problem: Analytical Projection Model for Non-Central Catadioptric Cameras with Quadric Mirrors

Structured Light 3D Scanning Under Global Illumination research figure

CVPR 2011 Structured Light 3D Scanning Under Global Illumination

Wide-angle light field capture using an array of spherical mirrors

SIGGRAPH Asia 2010 Axial-Cones: Modeling Spherical Catadioptric Cameras for Wide-Angle Light Field Rendering

Flexible Voxels for Motion Aware Videography research figure

ECCV 2010 Flexible Voxels for Motion Aware Videography

Image Invariants for Smooth Reflective Surfaces research figure

ECCV 2010 Image Invariants for Smooth Reflective Surfaces

Optimal Coded Sampling for Temporal Super-Resolution research figure

CVPR 2010 Optimal Coded Sampling for Temporal Super-Resolution

Axial Light Fields for Curved Mirrors: Reflect Your Perspective, Widen Your View research figure

CVPR 2010 Axial Light Fields for Curved Mirrors: Reflect Your Perspective, Widen Your View

Specular Surface Reconstruction from Sparse Reflection Correspondences research figure

CVPR 2010 Specular Surface Reconstruction from Sparse Reflection Correspondences

Reinterpretable Imager: Towards Variable Post Capture Space, Angle & Time Resolution in Photography research figure

Eurographics 2010 Reinterpretable Imager: Towards Variable Post Capture Space, Angle & Time Resolution in Photography

Invertible Motion Blur in Video research figure

SIGGRAPH 2009 Invertible Motion Blur in Video

Optimal Single Image Capture for Motion Deblurring research figure

CVPR 2009 Optimal Single Image Capture for Motion Deblurring

Coded Exposure Deblurring: Optimized Codes for PSF Estimation and Invertibility research figure

CVPR 2009 Coded Exposure Deblurring: Optimized Codes for PSF Estimation and Invertibility

Enforcing Integrability by Error Correction using L-1 Minimization research figure

CVPR 2009 Enforcing Integrability by Error Correction using L-1 Minimization

3D Pose Estimation and Segmentation using Specular Cues research figure

CVPR 2009 3D Pose Estimation and Segmentation using Specular Cues

Shield Fields: Modeling and Capturing 3D Occluders research figure

SIGGRAPH Asia 2008 Shield Fields: Modeling and Capturing 3D Occluders

Glare Aware Photography: 4D Ray Sampling for Reducing Glare Effects of Camera Lenses research figure

SIGGRAPH 2008 Glare Aware Photography: 4D Ray Sampling for Reducing Glare Effects of Camera Lenses

Detecting and Segmenting Un-occluded Items by Actively Casting Shadows research figure

ACCV 2007 Detecting and Segmenting Un-occluded Items by Actively Casting Shadows

Dappled Photography: Mask Enhanced Cameras for Heterodyned Light Fields and Coded Aperture Refocusing research figure

SIGGRAPH 2007 Dappled Photography: Mask Enhanced Cameras for Heterodyned Light Fields and Coded Aperture Refocusing

Resolving Objects at Higher Resolution from a Single Motion-Blurred Image research figure

CVPR 2007 Resolving Objects at Higher Resolution from a Single Motion-Blurred Image

Coded Exposure Photography: Motion Deblurring using Fluttered Shutter research figure

SIGGRAPH 2006 Coded Exposure Photography: Motion Deblurring using Fluttered Shutter

Edge Suppression by Gradient Field Transformation using Cross-Projection Tensors research figure

CVPR 2006 Edge Suppression by Gradient Field Transformation using Cross-Projection Tensors

What is the Range of Surface Reconstructions from a Gradient Field? research figure

ECCV 2006 What is the Range of Surface Reconstructions from a Gradient Field?

Removing Photography Artifacts using Gradient Projection and Flash-Exposure Sampling research figure

SIGGRAPH 2005 Removing Photography Artifacts using Gradient Projection and Flash-Exposure Sampling

An Algebraic Approach to Surface Reconstruction from Gradient Fields research figure

ICCV 2005 An Algebraic Approach to Surface Reconstruction from Gradient Fields

Why I want A Gradient Camera research figure

CVPR 2005 Why I want A Gradient Camera