Seeking AI Agent/LLM Application Job, focusing on Multi-Agent System, RAG Architecture and LLM-based Product Development
I am a postgraduate student majoring in Artificial Intelligence & Business Analytics at Lingnan University, with a solid background in Computer Science. I focus on AI Agent architecture design, RAG system development, and LLM application integration. I have rich experience in team project leadership and full-stack AI system development.
I am passionate about building practical LLM-based products and multi-agent systems, committed to transforming advanced AI technologies into scalable and valuable applications.
Problem: Develop an AI-driven obstacle avoidance game using reinforcement learning to realize automatic lane change and obstacle avoidance.
Data: Real-time game environment state data including lane position, obstacle distance and action feedback.
Approach: Implemented Q-Learning algorithm, designed reward function and epsilon-greedy exploration strategy.
Outcome: AI agent can independently complete obstacle avoidance tasks, stable performance in 3-lane game environment.
My Contribution: Independent development of full project including environment construction, agent design, training and deployment.
Problem: Users need a customized travel planner with budget, pace and preference constraints.
Data: Hong Kong attractions, food, accommodation and transportation dataset.
Approach: Built Planner + Critic multi-agent system based on Ollama & Mistral-7B.
Outcome: Realized intelligent itinerary generation with budget estimation, pace control and preference matching, stable multi-turn interaction.
My Contribution: As team leader, completed agent development, system integration and architecture design.
Problem: Students cannot quickly query university policies, courses and assignment information.
Data: University policy documents, course syllabi and academic notices.
Approach: Develop an intelligent Q&A system based on Retrieval-Augmented Generation (RAG) and Agentic architecture.
Outcome: Realized accurate intelligent Q&A for university information, reduced query costs and improved answer accuracy and reasoning ability.
My Contribution: Data collection, test case design, comparative experiments between AgenticRAG and RAG.
Problem: Build machine learning models to predict the survival probability of Titanic passengers and mine key factors affecting survival rate.
Data: Public Titanic passenger dataset including personal information, ticket information and survival status.
Approach: Data cleaning, EDA, feature engineering, classification model training.
Outcome: Completed the full data processing process, provided high-quality data support for model training and result analysis.
My Contribution: Responsible for data acquisition, data cleaning, preprocessing and Exploratory Data Analysis (EDA) of the Titanic dataset.
Problem: Complete image classification task on CIFAR-100 dataset, iterate and optimize from linear baseline model to residual CNN model.
Data: CIFAR-100 image dataset containing 100 categories of color image data.
Approach: Dataset splitting, data normalization, data augmentation, and linear baseline model training.
Outcome: Completed baseline model construction and data preprocessing, laid a foundation for the optimization of deep learning models.
My Contribution: Dataset processing, normalization, augmentation and linear baseline.
Download my full resume to learn more about my education, skills and projects.
Download CV (PDF)