Resume

Download Resume (PDF)

Current Position

• Research Scientist Intern     (Mar 202 – Present)
  MBZUAI, Institute of Foundation Models  Team: World Model
  • Working on multimodal LLM pretraining and post-training for vision alignment

🎓 Education

• Ph.D. Computer Engineering      (Oct. 2022 - Mar. 2026)
  University of California, Davis
  Dissertation Topic: Privacy-Preserving Computer Vision
  
• M.S. Computer Engineering      (Mar. 2021 - Mar. 2024)
  University of California, Davis
• B.S. Computer Engineering (Computer Science Minor)      (Sep. 2016 - Dec. 2020)
  University of California, Davis

💼 Prior Work Experience

• ML Engineer III Intern     (Sep 2025 – Present)
  Cisco Systems  Team: AI Defense
  • Exposed new vision-language vulnerability in multimodal prompt injection attacks
  • SFT of LLaVA-NeXT and Qwen2.5-VL for image safety assessment, boosting F1 score by ∼15%
• Applied Scientist Intern     (Apr 2025 – Aug 2025)
  Amazon  Team: Ring at Amazon Lab 126
  • Developed new architceture for multi-modal image retreival.
  • Created large scale multimodal fine grain retrieval dataset.
  • Proposed new pose-guidede feature disentanglement and dense disentangling loss.
  • Paper accepted at CVPR 2026
• Research Intern     (Jun 2023 – Sep 2023)
  Sony AI  Team: Privacy-Preserving Machine Learning
  • Developed and trained lightweight task-specific object detectors to detect PIIs to anonymize.
  • Adapted MobileNet-based architectures for on-camera detector inference.
  • Developed anonymization tool (mask, blur, inpaint, synthesize) for full body & face images.
  • Paper accepted at ICML 2024
• ML Research Engineer Intern    (Jul 2022 – Sep 2022)
  Sony  Team: Imaging & Sensing
  • Investigated Deep Learning (DL) based 3D reconstruction from images.
  • Tested and evaluated learning & non-learning based pipelines on custom datasets.
  • Modified and suggested suitable SOTA DL methods to integrate into existing pipeline.

📝 Select Publications

• Rendering-Refined Stable Diffusion for Privacy Compliant Synthetic Data
  K. Patwari*, D. Schneider*, X. Sun, C-N. Chuah, L. Lyu, V. Sharma*
  Under Submission   📄 Paper

• Empowering Source-Free Domain Adaptation via MLLM-Guided Reliability-Based Curriculum Learning
  K. Patwari*, D. Chen *, Z. Lai, X. Zhu, S. Cheung, C-N. Chuah
  WACV 2026   📄 Paper

• PerceptAnon: Exploring the Human Perception of Image Anonymization Beyond Pseudonymization for GDPR
  K. Patwari*, C-N. Chuah, L. Lyu, V. Sharma*
  ICML 2024   📄 Paper

🔧 Technical Skills

• Relevant Courses: Machine Learning, Vision and Language Research, ML Hardware, Image Processing
• Programming & Tools: Python, C/C++, CUDA, Docker, Git, Jupyter, Conda, Latex
• Programming/Frameworks: PyTorch, HuggingFace, OpenCilk, OpenCV, OpenMP, Scikit-Learn
• ML: Multimodal LLMs, Pruning, Adversarial Attacks, Diffusion, Domain Adaptation, Knowledge Distillation