Abhishek Kar

AI Researcher

San Francisco, California, United States8 yrs 3 mos experience

AI ML PractitionerAI Enabled

Key Highlights

Built ML-powered depth estimation for 1B+ devices.
Led ML function at Fyusion, driving innovation.
PhD in bridging 3D geometry and deep learning.

Stackforce AI infers this person is a leader in AR/XR and Machine Learning technologies.

Contact

kabhishek@meta.com LinkedIn

Skills

Core Skills

Machine LearningComputer Vision

Other Skills

3D Content Capture3D PerceptionAR Content DevelopmentAlgorithmsApplied ResearchArtificial IntelligenceCC++Computational PhotographyComputer ScienceData MiningData StructuresDeep LearningHands-on Technical LeadershipHuman Computer Interaction

About

Currently building the future of social connection at Meta. Before this, I built state-of-the-art ML-driven perception systems at Google powering the next generation AR/XR experiences. I enjoy "broad and deep" technical challenges and built performant real-world systems at the cutting edge of 3D perception, computational photography and computer graphics. Some of my highlights at Google included shipping ML-powered depth estimation scaled across 1B+ Android devices for improved realism in smartphone AR, depth-enabled computational photography features in Google Photos/Pixel/Android and (yet-to-be-disclosed) portions of the foundational mixed-reality technology stack. Prior to Google, I built and led the ML function at Fyusion, a spatial photography startup that pioneered casual capture, viewing and rich semantic understanding of light fields with the goal of democratizing 3D content capture and visualization using mobile devices. Fyusion was acquired by Cox Automotive in 2020. My formal training lies in bridging the worlds of 3D geometry and deep learning, graduating with a PhD from UC Berkeley with Prof. Jitendra Malik. You can find more about my research persona at https://abhishekkar.info

Experience

On deck

Founding Fellow, On Deck Deep Tech

Jan 2021 – Jan 2022 · 1 yr

Google

Research Scientist

Jan 2020 – Jan 2024 · 4 yrs · San Francisco Bay Area

Built and shipped 0-->1 systems at the intersection of ML-based 3D perception, neural rendering, computational photography and more at Google AR.
Highlights include ML-powered depth scaled across 1B+ Android devices for improving AR realism, depth-enabled computational photography features in Google Photos, Pixel and Android and (yet-to-be-disclosed) portions of the foundational mixed-reality technology stack.

Machine Learning3D PerceptionNeural RenderingComputational PhotographyComputer Vision

Fyusion, inc

2 roles

Head of Machine Learning

Promoted

Jan 2017 – Jan 2019 · 2 yrs

I started and led Fyusion's ML efforts with the goal of democratizing 3D/light-field captures using smartphones.
Highlights include:
Led the development of Fyusion’s flagship feature intelligent damage detection and mapping for cars which resulted in strategic investments and subsequent acquisition by Cox Automotive.
Technologies developed/shipped for automotive clients include deep-learning based object SLAM, automatic region-of-interest extraction, preset tagging and free-viewpoint rendering using local light field fusion.
Initiated and led academic collaborations. Filed a number of patents and co-authored articles in top-tier conferences.

Machine LearningDeep Learning3D Content CaptureComputer Vision

Visiting Research Scientist

May 2015 – Dec 2016 · 1 yr 7 mos

Developed solutions for User-generated AR content and real time ML inference on smartphones.

Machine LearningReal-Time InferenceAR Content DevelopmentComputer Vision

Microsoft

Research Intern

May 2011 – Jul 2011 · 2 mos · Redmond, WA

Undergraduate research intern in the Interactive Visual Media group under Dr. Michael Cohen and Dr. Neel Joshi. Worked on viewing large imagery on mobile devices.