Afsara Benazir

About Me

Hello! My name is Afsara Benazir. I am currently a fourth-year PhD student in the Department of Computer Science at the University of Virginia, advised by Prof. Felix Xiaozhu Lin. Prior to this, I received my B.Sc. in Computer Science from BUET (Bangladesh University of Engineering and Technology) in 2021. For more information, please check my CV.

My research interest lies in the domain of Systems X Machine Learning with a particular focus on

• ML acceleration
• On-Device AI and Foundation LLMs
• Speech: Execution of deep models on-device & user privacy
• Performance modeling across heterogenous hardware

I am motivated to build efficient AI systems to ensure optimal LLM serving. I build software stacks for optimizing performance for AI inference workloads on resource constrained devices. Currently my research explores scalable Mixture-of-Experts (MoE) architectures, LLM inference, resource efficiency, privacy, edge speech understanding and large-scale system benchmarking of quantized foundation models. I have build resource efficient, privacy preserving edge/cloud frameworks for speech understanding tasks on voice assistants.

I am seeking Spring/Summer AI/System Internship opportunities for 2026. Please feel free to reach out to me!

Publications

• [SIGMETRICS’26] Profiling Large Language Model Inference on Apple Silicon: A Quantization Perspective [PDF]
  Afsara Benazir, Felix Xiaozhu Lin

• [SEC’25] Privacy-Preserving Edge Speech Understanding with Tiny Foundation Models [PDF]
  Afsara Benazir, Felix Xiaozhu Lin

• [MobiSys’24] Speech Understanding on Tiny Devices with A Learning Cache [PDF]
  Afsara Benazir, Zhiming Xu, Felix Xiaozhu Lin

• [SOSP’25] A Journey of Modern OS Construction From boot to DOOM [PDF]
  Wonkyo Choe*, Rongxiang Wang*, Afsara Benazir*, Felix Xiaozhu Lin
  _{*Coprimary authors}

• Credibility assessment of User Generated health information of the Bengali language in micro blogging sites using NLP techniques and Machine Learning (WI-IAT’20) (Workshop paper) [pdf]
  Afsara Benazir, Sadia Sharmin

Poster

• Maximizing the Capabilities of Tiny Speech Foundation Models in a Privacy Preserving Manner (MobiCom’24)[pdf]
  Afsara Benazir and Felix Xiaozhu Lin

What’s New

• [10/2025] Our paper got accepted at SIGMETRICS’26!
• [10/2025] Our paper got accepted at the Symposium on Edge Computing (SEC’25)!
• [10/2025] Received faculty+audience choice award at the poster presentation session of UVa CS research symposium 2025! [poster]
• [05/2025] Officially a PhD candidate!
• [02/2025] Mentored 2 Charlottesville high school students in Spring’25 developing an engineering capstone project.
• [12/2024] Served as an artifact reviewer for PPoPP 2025
• [10/2024] Received a Student travel grant to attend MobiCom’24
• [10/2024] Student committee chair (lightning talk segment) at the 1st LLM workshop at UVA
• [10/2024] Our poster ‘Maximizing the Capabilities of Tiny Speech Foundation Models in a Privacy Preserving Manner’ is accepted at MobiCom’24 [pdf] [poster]
• [04/2024] Our paper titled ‘Speech Understanding on Tiny Devices with A Learning Cache’ is accepted at MobiSys’24)! [pdf]
• [04/2024] Presented our poster at CAPWIC’24 (ACM Capital Region Celebration of Women in Computing)
• [10/2023] Received faculty choice award at the poster presentation session of UVa CSGSG Research Symposium [poster]
• [10/2023] Flash Talk presentation at Link Lab Research Day [slides]
• [08/2022] Moved to Charlottesville and started my PhD journey at UVa CS!

Selected Awards

• 2022: Computer Science Scholar Fellowship, University of Virginia

Powered by Jekyll and Minimal Light theme.