Marcus's Projects
The APAC Datathon Report analyses Medicaid and Medicare payments for cancer treatments in the United States, assessing fairness, efficiency, and their impact on cancer mortality rates. It addresses outliers in payment amounts by state, explores the correlation between cancer mortality rates and Medicaid payments...etc
Using Librosa for audio file preprocessing, manipulating sound arrays, utilizing Waveplots and Spectrograms for insightful analysis, visually interpreting audio information, applying Fourier Transform for signal processing, employing signal filtering techniques like Windowing and Short-Time Fourier Transforms, and integrating OpenAI's Whisper AI.
Use predictive modelling to generate insights and recommendations for Grab, addressing delays in transport services and enhancing operational efficiency for an improved user experience.
Developed a context-aware query resolution system integrating LangChain with OpenAI's embeddings and ChatGPT, facilitating intuitive information retrieval and response generation through environmental variable management and relevance-based similarity search.
A dynamic showcase of my projects, skills, and experiences in web development. Built with HTML, CSS, and JavaScript, this website offers an interactive platform for visitors to explore my work and get in touch. Browse through projects, learn about my skills, and don't hesitate to provide feedback!
This project is a React blog application developed using Create React App. It includes features such as dynamic routing, data fetching from a JSON server, and integration with React Router. The project utilizes various React concepts including components, hooks, and state management. Check out the README for detailed instructions on installation:)
Incorporating computer vision for precise parcel sizing inputs from sellers improves efficiency in the seller-to-warehouse process, reducing rejected packages and disputes with NinjaVan.
The BestBy application utilizes a Telegram bot for receipt scanning via Optical Character Recognition (OCR) to determine Best By dates for purchased food items. This information is used to send users reminders, minimising food wastage.
Conducted a workshop for 30 NUS undergraduates on how to develop a CNN model and optimise it using KerasTuner
Hangman for TikTok is a project crafted during the TikTok Youth Camp 2022. Over two weeks of intensive training by TikTok engineers, we acquired skills in HTML, CSS, JavaScript, React, Android Development, and Git. The culmination of the camp involved creating a project that showcased the skills learned, resulting in the development of this.
Classification of 100k tweets using NLP techniques, evaluation of ML models like logistics regression, random forest and k-nearest-neighbours which includes hyperparameter optimisation using randomised search