Tom Lai

Using Prompting techniques to develop a chatbot system for helping customer questions.
The chatbot is designed to serve as a customer service assistant for the Learnspot website, a platform that aids students in preparing for tests and connecting with tutors
Using prompting techniques to assist with customer inquiries and prevent hallucinations. The assistant is programmed to respond in a friendly and helpful manner, providing concise answers and asking relevant follow-up.

To help rescue crews and retrieve the lost passengers in 2912.
I am challenged to predict which passengers were transported by the anomaly using records recovered from the spaceship’s damaged computer system.

My role in this project was to specialize in developing an effective recommender system.
The recommender system utilizes Filtering Collaboration to analyze user inputs and recommend 10 exercises based on ratings that closely match the user's preferences. The model is continuously improved through online batch updates based on user feedback and references.
Throughout the project, we emphasized the importance of PLESI (Privacy, Legal and Ethical, Security, and Internationalization) and addressed the issue of different genders using the app to ensure inclusivity.

The objective was to predict sales across all physical Pizza Hut stores using provided sales data.To achieve this, I leveraged my Python programming skills and applied Machine Learning techniques.
Developing and optimizing a time series model, utilizing TensorFlow's ANN model, to make accurate predictions. I then extended the model to forecast sales several months in advance. Additionally, I visualized the predictions to provide clear insights.
These pores result from a chemical reaction due to incorrect temperature and other parameters, leading to the formation of large holes in the metal blocks. This compromises their structural integrity, making them brittle and unstable, necessitating the cessation of the printing process to minimize costs.

Working with Dr. Muhammad Ansari on developing an algorithm model designed to automatically detect errors and halt the printing process of a 3D metal printer using YOLOv9 model.
My primary focus involves processing image labels using Connected Components Analysis. The dataset consists of images of metal blocks, with the goal of labeling the pores that appear during the printing process.
These pores result from a chemical reaction due to incorrect temperature and other parameters, leading to the formation of large holes in the metal blocks. This compromises their structural integrity, making them brittle and unstable, necessitating the cessation of the printing process to minimize costs.

Collaborating with a PhD candidate on a significant project involving the analysis of an eye-tracking dataset. My main role was to assist in deriving valuable insights from the dataset by categorizing fixations into "First Pass" (initial read) or "Second Pass" (subsequent read) sequences.
Realizing unusual points while analyzing the data, I developed an Anomaly Detection mechanism for the dataset where using the Local Outlier Factor (LOF) to identify and remove outliers, establish criterias for accurately determining when a reader transitioned to the next line of text. These contributions notably improved the robustness and accuracy of the data analysis process.
The project required meticulous work to understand readers' eye movements and reading patterns. The dataset encompassed approximately 20 features, tracking various aspects of readers' eye movements, including fixation coordinates, durations, and directions.
Reference: Minh Huyen's post

Hi