Conducted multiple HPC resilience studies during PhD, including:
- Developed the HAPPA platform for SDC prediction with MSE of 0.078, outperforming existing models by 30%
- Performed semantic-based loop resilience analysis across the '13 dwarfs of parallelism'
- Evaluated LLM capabilities in understanding Intermediate Representations (IRs) for low-level program analysis
Pioneered empirical studies on applying LLMs to compiler design and program analysis, benchmarking GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama on tasks such as CFG reconstruction, IR decompilation, code summarization, and execution reasoning
Actively exploring interdisciplinary AI applications in healthcare, including medical data pattern recognition, interpretable AI, and robustness of healthcare AI systems