Rachel

My MSc Project employed Python and the Apify API to meticulously extract and preprocess recipe data from AllRecipes, ensuring data accuracy and quality. Leveraged FooDB to identify the most prevalent base ingredients within the acquired dataset. Utilized Word2Vec to produce embeddings for each ingredient, facilitating the discovery of similarities and relationships between ingredients. Developed a robust Neo4j graph database using Cypher, Py2Neo, and the GDS Library, enabling efficient data storage and retrieval. Implemented a Graph Neural Network (GNN) to perform advanced machine learning tasks, capitalizing on the rich data structure of the Neo4j database. Employed GPT-3.5, an OpenAI API, to generate innovative flavor profiles and ingredient combinations by leveraging the relationships and knowledge within the graph database.

Designed and implemented a multi-view pose estimation model to enable precise human action recognition within a 3D environment. • Leveraged the capabilities of multiple Nvidia Jetson TX2 Developer Kits to establish a dynamic camera array for continuous subject tracking. • Utilized a tech stack comprising Python, Jupyter Notebook, and TensorFlow to develop a robust application for real-time subject tracking and action recognition