Muhammad Nahid

I am a final year Computer Science undergraduate looking to gain experience in the world of Machine Learning and Software Development. I aim to bridge the 2 worlds together to create products that are useful and bring benefit to the community. I believe no problem is too hard to solve with the right amount of focus and determination.

ClaryFy NoteTaking WebApp (https://claryfy.vercel.app/login)

• Integrated GPT-4 Image Recognition API to convert handwritten notes into digital text, achieving 83% recognition accuracy and improving study productivity for users.
• Developed a robust back-end system using MongoDB for scalable and efficient API calls.
• Designed and implemented a user-friendly UI/UX using TypeScript and React.

Nutri-Vision MobileApp (https://youtu.be/K2hoFSBUiCc)

• Developed a food image recognition app using GPT-4 API, providing users with instant nutritional information at 99% test accuracy. Integrated GPT-4 API for image recognition and Calorie-Ninja for nutritional data, enhancing user accessibility to dietary information.
• Designed a user-friendly logging feature to track meals and monitor dietary progress, supporting users in achieving personalised health goals.

MDP Image Recognition

• Collected and prepared a diverse dataset using RoboFlow for training a YOLO-v5 image recognition model.
• Developed an image recognition model using YOLO-v5, achieving over 97% accuracy on the competition dataset.
• Deployed the trained model on a Raspberry Pi using an inference server for real-time object detection.
• Detected all images during the competition, contributing to a top 8 finish out of 54 teams in the NTU CS cohort.

Flowers102 ImageRecognition

• Experimented with multiple deep learning models using PyTorch, achieving a test accuracy of 0.855 on the Oxford Flowers 102 dataset.
• Designed, trained, and evaluated CNN, CNN-Transformer hybrid, ViT, and MLAN Xception models to benchmark performance and optimise accuracy.