CV

Contact Information

Name Danial Samadi Vahdati
Professional Title PhD Candidate, Electrical & Computer Engineering
Email danielsvahdati@gmail.com

Professional Summary

AI security researcher and PhD candidate specializing in deepfake detection, synthetic media forensics, and adversarial ML. 5+ years of research experience with 5 first-author publications at top-tier venues (NeurIPS, CVPR) and 60+ citations. Developed real-time detection systems in collaboration with NVIDIA Research, securing a $150K research gift. Contributed to $3M+ in federally funded research (NSF, DARPA, NIST).

Experience

  • 2021 - Present

    Philadelphia, PA

    Graduate Research Assistant
    Drexel University — Multimedia & Information Security Lab
    • Designed first enrollment-free defense against identity puppeteering in AI videoconferencing — 97.7% AUC at 75 FPS, 46% error reduction over prior SOTA. Published at NeurIPS 2025.
    • Engineered end-to-end AI-generated video detection pipeline achieving 99%+ accuracy across Sora, Runway AI, Stable Video Diffusion, and NeRF-based systems. Published at CVPR 2024.
    • Introduced virtual generator technique to improve synthetic video detector generalization via parametric forensic microstructure synthesis. Under review at IHMMSEC 2026.
    • Built and released 8M+ frame benchmark dataset on Hugging Face (400+ downloads, adopted by external research groups).
    • Contributed to $3M+ in federally funded research (NSF, DARPA, NIST).
    • Peer reviewer: 50+ reviews for NeurIPS, CVPR, ICCV, IEEE TIFS, IHMMSEC, ICIP.
    • Mentored 2 junior PhD researchers and 1 undergraduate student.
  • 2024 - Present

    Remote

    Research Collaborator
    NVIDIA Research
    • Active research collaboration on deepfake detection and synthetic media forensics.
    • Co-authored NeurIPS 2025 paper with NVIDIA researchers (Ekta Prashnani, Koki Nagano, Orazio Gallo).
    • Collaboration secured $150K NVIDIA research gift and 1 year of dedicated compute resources.

Education

  • 2021 - 2026

    Philadelphia, PA

    PhD
    Drexel University
    Electrical & Computer Engineering
    • GPA: 3.85
    • Advisor: Prof. Matthew C. Stamm
    • Multimedia & Information Security Lab (MISL)
    • Research focus: Deepfake detection, synthetic media forensics, AI security, adversarial ML
  • 2022 - 2024

    Philadelphia, PA

    MS
    Drexel University
    Electrical & Computer Engineering
  • 2016 - 2020

    Qazvin, Iran

    BS
    Imam Khomeini International University
    Electrical Engineering
    • Top 1% of graduating class

Publications

  •
    Unmasking Puppeteers: Leveraging Biometric Leakage to Expose Impersonation in AI-Based Videoconferencing
  •
    Seeing the Unseen: Enhancing Synthetic Video Detector Transferability via Virtual Generators
  •
    Beyond Deepfake Images: Detecting AI-Generated Videos
  •
    Defending Low-Bandwidth Talking Head Videoconferencing Systems From Real-Time Puppeteering Attacks
  •
    Detecting GAN-Generated Synthetic Images Using Semantic Inconsistencies

Skills

Research Areas: Deepfake detection, AI-generated media forensics, Adversarial ML, Synthetic media authentication, Real-time vision systems, Contrastive learning, Generative AI security
Programming Languages: Python, C++, C, MATLAB, Bash
ML/DL Frameworks: PyTorch, PyTorch Lightning, TensorFlow, CUDA, HuggingFace Transformers
Deep Learning: CNNs, Transformers, GANs, Diffusion Models, NeRFs, Self-Supervised Learning, Contrastive Learning, Vision-Language Models
Computer Vision: OpenCV, MediaPipe, Video forensics, Image forensics, Object detection, Temporal modeling
Generative AI: Stable Diffusion, ComfyUI, Prompt engineering, LLM evaluation, GPT-4o, Gemini, Synthetic data generation
MLOps & Tools: Git, Docker, AWS, Weights & Biases, Linux, SLURM, FastAPI, HuggingFace Hub

Service

Peer Reviewer: 50+ reviews for NeurIPS, CVPR, ICCV, IEEE TIFS, IHMMSEC, ICIP
Mentorship: Onboarded and mentored 1 junior PhD researchers and 1 undergraduate student
Open-Source: Released large-scale synthetic media benchmark dataset on Hugging Face (400+ downloads)
Professional Membership: IEEE Student Member