Experience

Argonne National Laboratory

Visiting Graduate Researcher · Lemont, Illinois

Developing hardware–software co-designed scientific data-compression systems for the Advanced Photon Source.

• Designed an end-to-end FPGA/CPU compression framework achieving up to 19.8× cycle speedup over matched software and 46–65% lower end-to-end workflow time.
• Developed a modular Chisel/RTL compression library supporting LZ4, Snappy, StreamVByte, SZ3, SZp, SZx, zfp, and extensible new designs.
• Advised by Dr. Sheng Di in collaboration with Dr. Guanpeng Li.

University of Florida

Graduate Research Assistant · Dependable Systems Lab · Gainesville, Florida

Characterizing the resilience of LLM inference, pretraining, and parameter-efficient fine-tuning to transient hardware faults.

• Developed TRACE, a runtime silent-error detector achieving 95.3% average detection coverage across 15 LLMs and 3 datasets with 0.56% runtime overhead; paper accepted at SC26.
• Characterizing fault resilience across LLM inference, pretraining, and models fine-tuned with LoRA, QLoRA, LoRA-FA, and DoRA.
• Advised by Dr. Guanpeng Li.

The University of Iowa

Graduate Research Assistant · Iowa City, Iowa

Studied compiler- and runtime-based silent-error detection for scientific and HPC applications.

• Developed an LLVM-based silent-error detector using selective monitoring of informative program locations, achieving 86.19% average detection coverage at 4.90% runtime overhead across 15 benchmarks.
• Compared compiler- and runtime-based error-detection techniques across scientific workloads, analyzing their behavior under different fault-propagation patterns and quantifying trade-offs between detection coverage and runtime overhead.

The University of Iowa

Teaching Assistant · Database Systems

Led labs and office hours while mentoring more than 30 students through database projects, query debugging, and relational schema design.

Lahore University of Management Sciences

Teaching Assistant · Data Structures

Guided more than 100 students through tutorials and office hours in programming, data structures, algorithms, and debugging.