Rana Muhammad Shahroz Khan

I am a first year PhD student at UNC Chapel Hill where I work on Efficient ML and NLP. I am grateful to be advised by Dr. Tianlong Chen. Before that I did my Bachelors in Computer Science and Mathematics from Vanderbilt University, where I was advised by Dr. Soheil Kolouri and worked in the intersection of Computer Vision and Optimal Transport.

During the summer of 2023, I was a Research Intern at Lawrence Livermore National Laboratory in the Machine Intelligence Group working under the guidance of Dr. Jay Thiagarajan, Dr. Shusen Liu and Dr. Rushil Anirudh on Bayesian Optimization for Uncertainty Quantification. In the Summer of 2024, I also had the pleasure of working as a ML Research Intern at HighArc under the supervision of Dr. Ardavan Bidgoli and Dr. Manuel Ladron de Guevara.

Email  /  GitHub  /  Google Scholar  /  LinkedIn  /  CV [01/2025]

profile photo

Updates

  • [May 2025] Our preprint "Agents Under Siege: Breaking Pragmatic Multi-Agent LLM Systems with Optimized Prompt Attacks" have been accepted to ACL 2025's Main Track !
  • [April 2025] Excited to give a talk on our preprint "Agents Under Siege" at ResearchTrend.AI
  • [April 2025] Two new preprints on: (1) Conditional LoRA Generation and (2) Safety of Multi-agent Systems.
  • [March 2025] I will join AWS Bedrock for Fall 2025.
  • [March 2025] I will be joining Cisco for Summer 2025.
  • Research

    I'm interested in NLP and Generative AI, with hopes of making it efficient.

    project image

    ORAL: Prompting Your Large-Scale LoRAs via Conditional Recurrent Diffusion


    Rana Muhammad Shahroz Khan, Dongwen Tang, Pingzhi Li, Kai Wang, Tianlong Chen
    Preprint, 2025
    paper /

    ORAL leverages conditional recurrent diffusion to instantly craft task‑tuned LoRA adapters that scale to billion‑parameter models, remain compatible with future model updates, and rival—or beat—full fine‑tuning accuracy.

    project image

    Agents Under Siege: Breaking Pragmatic Multi-Agent LLM Systems with Optimized Prompt Attacks


    Rana Muhammad Shahroz Khan, Zhen Tan, Sukwon Yun, Charles Fleming, Tianlong Chen
    Association for Computational Linguistics (ACL), 2025
    paper /

    Agents Under Siege introduces a flow‑optimized, order‑agnostic prompt attack that stealthily threads limited‑bandwidth, high‑latency multi‑agent LLM networks to jailbreak target models, overwhelming modern safety guards and boosting attack success up to seven‑fold.

    project image

    PortLLM: Personalizing Evolving Large Language Models with Training-Free and Portable Model Patches


    Rana Muhammad Shahroz Khan, Pingzhi Li*, Sukwon Yun*, Zhenyu Wang, Shahriar Nirjon, Chau-Wai Wong, Tianlong Chen
    International Conference on Learning Representations (ICLR), 2025
    paper /

    We present PORTLLM, a training-free framework that enables seamless knowledge transfer across evolving LLMs, achieving LoRA-level performance with significantly lower computational costs.

    project image

    LLMGeo: Benchmarking Large Language Models on Image Geolocation In-the-wild


    Zhiqiang Wang, Dejia Xu, Rana Muhammad Shahroz Khan, Yanbin Lin, Zhiwen Fan, Xingquan Zhu
    Conference on Computer Vision and Pattern Recognition, Workshop on Computer Vision in the Wild (CVPRW), 2024
    paper / code /

    We evaluate multimodal LLMs for image geolocation using a new dataset, showing that fine-tuning helps open-source models approach closed-source performance.

    project image

    PRANC: Pseudo RAndom Networks for Compacting Deep Models


    Parsa Nooralinejad, Ali Abbasi, Rana Muhammad Shahroz Khan*, Soroush Abbasi Koohpayegani*, Kossar Pourahmadi Meibodi*, Soheil Kolouri, Hamed Pirsiavash;
    International Conference on Computer Vision (ICCV), 2023
    paper / code /

    We propose PRANC, a framework that reparametrizes deep models as a linear combination of frozen random networks, enabling extreme compression, efficient storage, and memory-efficient inference.

    project image

    Linear Optimal Partial Transport Embedding


    Yikun Bai, Ivan Vladimir Medri, Rocio Diaz Martin, Rana Shahroz, Soheil Kolouri
    International Conference on Machine Learning (ICML), 2023
    paper / code /

    We propose the Linear Optimal Partial Transport (LOPT) embedding, enabling faster OPT distance computation and demonstrating its effectiveness in point-cloud interpolation and PCA analysis.





    Design and source code from Jon Barron's website