CV

Education

Aug. 2024 - present: Embodied AI + Model Compression
- Compressing Vision Transformer models using real post-training quantization, reducing precision from Float16 to Int8 by incorporating CUDA extensions to achieve 2× faster model inference.
- Deploying quantized vision models on Limo Pro cars, enhancing robustness and efficiency in multi-robot applications through a tailored W8A8 quantization method.
Dec. 2023 - May. 2024: Agentic AI + RAG
- Developed a medical chatbot that answers concerns by fetching the latest theses and utilizing RAG search for accurate information.
Sep. 2023 - Dec. 2023: Trustworthy AI
- Conducted research on interpretability, ethics, and bias in generative models deployed on a K-12 learning assistance platform.

May 2024 - Aug 2024: Development Intern at Prometheus Group, Raleigh
- Addressed LLM token length limitations by implementing strategic planning optimizations, enabling the model to efficiently query terabytes of SAP data.
- Developed a multi-modal agentic LLM for precise issue diagnosis, integrating image analysis, time-series data, and discussions from issue report forums.
- Accelerated an LLM-based chatbot, achieving an 87.5% reduction in runtime for answering frequently asked questions.
Sep. 2020 - Aug. 2022: Junior Data Scientist at Shinewave (Subsidiary of ASUS), Taipei
Responsible for anomaly detection and root cause analysis for fuse production line * Conducted anomaly detection and root cause analysis for developing early warning systems in real-time production monitoring system to detect defectives
- Developed the end-to-end analytic product development process, from data extraction in monitoring system, streaming data pipelines design, customized dashboard for visualization, and data narrative writing.
- Analyzed abnormal data from monitoring systems to identify human errors, designing visualization reports to provide actionable insights for factory managers in manufacturing industry.

Reviewer