CV
Contact Information
| Name | Xinyu Gong |
| Professional Title | Applied Scientist |
| neoxygong@gmail.com |
Professional Summary
Applied Scientist working on generative models for image/video.
Experience
-
2026 - present San Jose, CA
Applied Scientist
Adobe Firefly
-
2023 - 2026 San Jose, CA
Machine Learning Engineer
TikTok
- Improved text-to-image models using supervised fine-tuning (SFT) and curated training datasets, enhancing prompt adherence and visual quality.
- Performed post-training on image-to-video models with SFT and RLHF, improving temporal consistency and motion realism.
- Curated preference datasets and trained reward models to support RLHF and strengthen model alignment.
-
2022 - 2022 Sunnyvale, CA
Research Intern
Meta Reality Labs
- Proposed and studied Multimodal Generalization (MMG) for robustness when certain modalities are limited or missing.
- Introduced MMG-Ego4D dataset and built baseline models using contrastive learning and cross-modality alignment.
-
2021 - 2022 Austin, TX
Research Intern
PicsArt AI
- Built a few-shot GAN for efficient incremental class learning using a hypernetwork and data augmentation.
-
2021 - 2021 Menlo Park, CA
Research Intern
Facebook AI
- Designed an incremental few-shot object detection approach with weakly-supervised augmentation and compact architecture design.
-
2020 - 2020 Menlo Park, CA
Research Intern
Facebook AI
- Developed an efficiency-oriented NAS method for video action recognition with a multivariate two-stream search space and progressive search.
Education
-
2021 - 2023 Austin, TX
Ph.D.
The University of Texas at Austin
Electrical and Computer Engineering
- Advised by Dr. Atlas Wang.
-
2018 - 2021 College Station, TX
Ph.D.
Texas A&M University
Computer Science
- Advised by Dr. Atlas Wang.
-
2014 - 2018 Chengdu, China
B.E.
University of Electronic Science and Technology of China
Computer Science
Awards
-
2024 IEEE Signal Processing Society (SPS) Young Author Best Paper Award
Service
- Reviewer: AAAI, CVPR, ICCV, ICML, NeurIPS, IJCV
- Co-organizer: ML4Wireless workshop at ICML 2025
Skills
Programming: Python, Bash, Matlab
ML Frameworks: PyTorch, TensorFlow, Keras
Interests
Machine Learning: Few-shot Learning, Neural Architecture Search, Generative Models
Applications: Image/Video Generation