Short Bio
Currently, I work in stealth mode.
My passion lies in exploring the boundaries of AI models with human-like visual & linguistic (and/or emotional) capabilities.
My area of focus is on designing deep learning models for multi-modal data. Specifically, my typical methods use natural language to emphasize semantic differences or the emotional characteristics of depicted objects in visual data.
I recently spent a great year being a Research Scientist working for the Creative Vision team of Snap Research.
I received my Ph.D. degree from the Department of Computer Science Department of Stanford University for my work done with the Geometric Computing Lab under the supervision of Leo Guibas.
In the past, I interned at the Facebook AI Research Lab in Menlo Park and, before that, at Autodesk Research in San Francisco. A few years back, I was a research assistant in the Haussler Lab at UCSC and an Erasmus scholar at the Max Planck Institute for Intelligent Systems in Tüebingen, DE.
Outreach
On weekends, depending on availability, I voluntarily host office hours for students (especially underrepresented groups and junior students) who want to get into or dive deeper into the fields of Machine Learning, Computer Vision, or NLP. Each slot is 20-minutes long. If you want to schedule a meeting, please fill out this questionnaire.
Latest News
June 2023: I have two first-author papers appearing in CVPR-2023, Vancouver. Affection and ShapeTalk. A big shout-out to all my co-authors who have made these works a reality shared at the top-tier level of our research community.
April 2023: Our third workshop at the intersection of 3D scenes and natural language for common objects (L3DS) will be part of the ICCV-2023 workshop series.
December 2022: My first last-author paper is now a reality (ScanEnts3D). In this work, we exploit dense correspondences between 3D objects in scenes and referential language and improve the SoTA in two essential multi-modal tasks in 3D scenes. Big shout-out to my amazing intern Ahmed Abdelreheem for all his hard work.
November 2022: LADIS, a rigorous framework for improving shape editing models of 3D objects via language, will appear in the findings of EMNLP-2022.
October 2022: A big milestone in my "quest" of developing more emotionally aware, and human-centric AI is achieved. Affection is now on arXiv.
April 2022: Dance2Music-GAN, a practical adversarial multi-modal framework that generates complex musical samples conditioned on dance videos, will appear at ECCV-2022.
March 2022: Our second workshop at the intersection of 3D scenes and natural language for common objects (L3DS), will be part of the ECCV-2022 workshop series. We are looking forward to a happy reunion and passionate, productive discussions.
March 2022: NeROIC, a novel object acquisition framework which exploits and extends radiance fields to capture high-quality 3D objects from online image collections, will appear at SIGGRAPH-2022.
March 2022: PartGlot, which opens the door for the automatic recognition and localization of shape-parts via referential language alone, will appear with an oral presentation in CVPR-2022.
November 2021: Gave a talk describing recent trends on Affective Deep Learning at Stanford's STATS 281 Statistical Analysis of Fine Art.
October 2021: I am excited to start my new role as a Research Scientist for the Creative Vision of SNAP Research.
May 2021: The content of our CVPR-21 workshop L3DS concerning language and 3D scenes has been finalized! Among others, we will host a benchmark challenge for ReferIt3D: (here).
March 2021: ArtEmis keeps growing. Now it is featured in Forbes Science.
March 2021: I successfully defended my Ph.D. Thesis titled "Learning to Generate and Differentiate 3D Objects Using Geometry & Language".
March 2021: I will give a lightning talk on "Art and AI" during HAI’s Intelligence Augmentation: AI Empowering People to Solve Global Challenges.
March 2021: Our work ArtEmis: Affective Language for Visual Art is provisionally accepted as an Oral presentation in CVPR-2021.
February 2021: Our recent arXiv report (ArtEmis) attracted some media attention: New Scientist, HAI, MarkTechPost, KCBS-Radio (want to hear me talk about it? check the short interview below):
February 2021: I will co-organize the 1st Workshop on Language for 3D Scenes in CVPR 2021. We hope to spark new interest in this emerging area!
February 2021: I am initiating this "News" section. My intention is to give the gist of my (primarily) professional updates to visitors.