Research Assistant Computer Science
Resume:
Melissa Liu
EDUCATION
University of California, Davis Ph.D. Computer Science GPA: 3.82 2027 Expected Graduation Robert D. Clark Honors College, University of Oregon B.S. Math, Economics GPA: 3.73 2019 Presidential Scholar, Wayne Morse Scholar, University Student Government, Mock Trial EXPERIENCE & RESEARCH
UC Davis, Automated Reasoning Group — Graduate Research Assistant March 2024 – present
• Improving the safety and reliability of deep learning models – currently leveraging Large Language Models to address floating-point arithmetic errors for a widely-used scientific HPC application.
• Evaluating Large Language Models in Floating-point Understanding pending CAV 25 o Presents FPEval, a benchmark suite that evaluates LLMs’ floating-point understanding. o FPEval demonstrates that the state-of-the-art LLMs are not reliable at writing or rewriting floating-point programs and can incorrectly recall algorithms and formulas or introduce numerical inaccuracies in existing programs that are difficult to discern without additional verification.
UC Davis, Computer Vision Lab — Graduate Research Assistant May 2023 – March 2024
• Towards a Robust Ingredient Prediction Network 2024 with USDA o Presents Weighted Layer-wise Ensemble of Different Views (W-LEVI), a novel fine-tuning method which allows a model to adapt to the new fine-tuned dataset while enhancing its ability to generalize to unseen data.
o W-LEVI beats the state-of-the-art fine-tuning method on the fine-tuned dataset Recipe1M by 3% and beats zero-shot performance on SNAPMe by 4%, an out-of-distribution food images dataset.
• Effective Adversarial Data Augmentation for BERT Models 2023 o Introduces method for generating and augmenting natural language adversarial data at scale, improving BERT language model accuracy by 6.9% on sentiment analysis. UC Davis Computer Science Department — Teaching Assistant September 2023 – present
• Teaching assistant for graduate level Computer Vision and Advanced Deep Learning courses, leading weekly discussions and cover lectures ad hoc. Arnerich Massena — Research Analyst
October 2019 – June 2022
• Designed and implemented an automated macroeconomic data collection for client portfolio risk- management with Python.
• Rebuilt SQL databases and optimized data pipelines using Power BI and Microsoft SQL Server. SKILLS
Programming Languages — Python, Java, C++, R
Tools — PyTorch, TensorFlow, Excel, Bloomberg Terminal, FactSet, Morningstar Direct Languages – Mandarin (professional proficiency), French (professional proficiency)
*******@*****.***
linkedin.com/mliu720
Contact this candidate