Hi, I am Fisokuhle Thusini.

A Streamlit app that is a continuation of the FGT Chat app where the GPT-3.5 model is used alongside the Llama2-7b model to showcase the capabilities of the models and the benefits that each provide, where a user pays for what they use in the GPT-3.5 model, reducing costs and they can have the Llam2-7b model self hosted, providing better security.

An iOS developed using iOS features including computer vision for personalised workouts, where you can see the percentage of your workout compared to how the workout is commonly done. There are also features that connect to the Apple Watch where a person can view their different health metrics such a steps and their average heart rate, as well as being able to to schedule workouts on the app.

A PWA used to collaboratively learn in groups using LLMs together to improve knowledge and learn in teams, the GPT-3.5 model is used to provide a chat application using private data stored in a private Pinecone vector database. The project was continued with a ChatGPT plugin for those with ChatGPT Plus to use the model via ChatGPT or using the FGT Chat PWA. Learn more...

There is a voice option to interact with the GPT-3.5 model, a recorded voice note is transcribed and sent to the model.

Data from the OpenAQ API visualized using streamlit to diplay the pollutants in different locations in real-time. Using that information to see trends in air quality around the world and identify any lack of information reporting.

A website created using PHP with the Google Books API that provides an opportunity to book lovers and students to search for books in Google Books, get previews and be provided with options on where to buy these books. With a discussion board to be able to share books or discuss specific books that they are currently reading.

A web app created with React JS, Node JS, and Google Firebase. The web app uses Firebase authentication to log in using email or with a Google account, then the chats are stored in the Firebase database which is a NoSQL database, with the files shared stored in the Firebase storage and those files can be shared and viewed in the web app.

A web app developed with a colleague using Python Flask, HTML, and CSS used to manage stock and staff with different access to the system. The application is designed with a user friendly UI, using PostgreSQL to manage and store the data. With the AWS EC2 instance used to host the web application.

A web app developed with Streamlit that displays original stock trends compared to predicted stock trends that have been predicted by a model trained using the LSTM (Long Short Term Memory) neural network to learn long term dependencies.

Prevention is better than cure, a computer vision project using OpenCV, CVZone and Pyserial to detect construction Hard Hats and Eyeware to improve safety in the workplace. With voice prompts, Arduino outputs, and video recording when the Hard Hat is off. With a user interface for better management of the system.