Computer Science and Engineering
Michigan State University
Dissertation: Proactive Schemes: Adversarial Attacks for Social Good
CV
Vishal Asnani / Research Scientist
Trustworthy AI. Traceable media. Proactive safeguards.
A snapshot of my research, experience, and technical background. For a concise application-ready version, download my two-page resume.
Research focus
Trustworthy generative AI; image, video, and audio watermarking; content authenticity; media provenance; causal attribution; manipulation detection and localization; model reverse engineering.
Experience
PresentResearch Scientist
Adobe Research · San Jose, CA
Content authenticity, responsible AI, and durable video Content Credentials, with current work on video and audio watermarking.
Mentored interns on TokenTrace (CVPR 2026); co-developed object-level watermarking (ICCV 2025) and NRGMark (WACV 2026).
Research Scientist Intern
Adobe Research · May - November, each year
Developed CustomMark (ICCV 2025 APAI Workshop) and ProMark (CVPR 2024) for proactive attribution in diffusion models.
Doctoral Research
Michigan State University · Advisor: Prof. Xiaoming Liu
Proactive manipulation detection and localization, generative-model parsing, and proactive object detection. Research collaborations with Meta AI and DEVCOM Army Research.
Education
Electronics and Instrumentation Engineering
BITS Pilani
Technical expertise
- Languages
- Python, C/C++, SQL
- ML & vision
- PyTorch, PyTorch Lightning, TensorFlow, NumPy, scikit-learn, OpenCV, CUDA
- Research methods
- Diffusion-model customization, textual inversion, token embeddings, encoder-decoder architectures, vision-language models, LoRA
- Cloud
- AWS, Google Cloud Storage
Service and communication
- Co-organizer, Foundations and Frontiers of Watermarking, CVPR 2026.
- Co-organizer, Reverse Engineering of Deceptions, CVPR 2023.
- Reviewer for computer-vision and machine-learning conferences and journals.
- Research presentations at Scale AI.
See the publication list and talks and tutorials for details.