About
I recently graduated with an M.S. in Computer Science (Machine Learning Specialization) from Georgia Tech. During my time at Tech, I conducted research as a Ph.D. student in the Data to Insights Lab under the guidance of Professor Kexin Rong, developing interpretable systems for object detection and video querying. Prior to this, I collaborated with Stefan Savage, Geoffrey M. Voelker, and Ariana Mirian at UC San Diego on various Internet measurement projects.
My interests lie at the intersection of data, security, and machine learning systems. I have worked on projects ranging from efficient LLM post-training and quantization to building multimodal video search engines.
Curriculum Vitae
Projects
Resource-Efficient LLM Post-Training
Accelerated post-training for Qwen-4B using Unsloth (LoRA/quantization) for RLHF on a single 24GB GPU. Implemented semantic data filtering and experimented with GRPO, DrGRPO, and DAPO algorithms.
LLM Based Video Clipper
Developed a pipeline using Whisper and Gemini Flash 2.0 to identify and stitch interesting video clips with transitions and subtitles.
Papers
VCR: Interpretable and interactive debugging of object detection models with visual concepts
Jie Jeff Xu, Saahir Dhanani, Jorge Piazentin Ono, Wenbin He, Liu Ren, Kexin Rong
Demonstration of VCR: A Tabular Data Slicing Approach to Understanding Object Detection Model Performance
Jie Jeff Xu, Saahir Dhanani, Jorge Piazentin Ono, Wenbin He, Liu Ren, Kexin Rong
Rethinking Similarity Search: Embracing Smarter Mechanisms over Smarter Data
Renzhi Wu, Jingfan Meng, Jie Jeff Xu, Huayi Wang, Kexin Rong
Hot Pixels: Frequency, Power, and Temperature Attacks on GPUs and Arm SoCs
Hritvik Taneja, Jason Kim, Jie Jeff Xu, Stephan van Schaik, Daniel Genkin and Yuval Yarom
Contact
Email: jeffxu2018 (AT) gmail.com