About me

Research Interest: Natural Language Processing (NLP), Large Language Models (LLMs), and Trustworthy Artificial Intelligence

I am a last-year PhD candidate specializing in Natural Language Processing (NLP), Trustworthy Artificial Intelligence (AI), and Machine Learning (ML/LLMs), with a strong background in various programming languages. Highly motivated, I have contributed to several research projects and publications over the years, showcasing my expertise and dedication to the field. With 8 years of expertise in NLP models and tasks, over 4 years of experience in secure AI, and robust PLMs & LLMs, I am a Robust Machine Learning Researcher proficient in programming languages and algorithm design, with over 10 years of experience. I am currently conducting research aimed at enhancing the robustness and efficiency of Pre-trained Language Models (PLMs) for adversarial NLP applications, and I am scheduled to defend my PhD dissertation in the next few weeks. Actively seeking full-time positions as an NLP (preferably Generative AI) Scientist/Engineer or Machine Learning Researcher/Engineer, I am eager to contribute to the further development of robust and efficient PLMs/LLMs for real-world applications.

As part of my previous practical work, I have undertaken a variety of projects that showcase my expertise in natural language processing and machine learning, specifically in large language models. I participated in an industrial project aimed at detecting various types of plagiarism (e.g., paraphrases and semantic plagiarism) within a corpus of 2.5 million papers. Furthermore, I played an integral role in developing a sophisticated text alignment framework, where I explored fundamental data mining techniques to precisely detect text reuse segments. I fine-tuned a pre-trained language model (T5) for text summarization using the Prefix-Tuning technique on the CNN/Daily Mail dataset. Additionally, I built Retrieval-Augmented Generation (RAG) systems using LlamaIndex, which involved loading data from various sources, creating vector indexes, querying the indexes, and evaluating the system’s performance. I also explored different prompt engineering techniques, such as Few-shot, Zero-shot, and Chain-of-Thought, to build a useful LLM application. In the realm of sentiment analysis, I created a multi-modal sentiment analysis model for movie reviews using RNN/LSTM, deployed it on AWS, managed the codebase with Git, and utilized multi-modal analysis. Furthermore, I extended a conversational AI assistant to handle customer inquiries and provide support for a hypothetical e-commerce company. Lastly, I developed an intelligent document processing system using Azure AI and Machine Learning services to extract relevant information from scanned documents, classify them based on their content, and provide insightful analysis.

News