AP

Designed and developed a human-like robotic system using Softbank Robotics' Pepper robot, capable of delivering interactive lectures. Utilised Large Language Models for slide and script generation, DNNs for speech synthesis (TTS) and recognition (STT), and vector-based databases for accurate Q&A retrieval. (ROS, Python)

Designed and built a Machine Learning system capable of predicting tennis serve outcomes using custom-built hardware. Collaborated closely with a multidisciplinary team to integrate hardware and software components, ensuring seamless data collection and analysis. Developed software for data acquisition and real-time processing. (Python, C++)

Analysed drivers performance and produced infographics as well as maps for better data visualisation using telemetry recordings from live GP races. (Python)

Implemented Deep Learning models (LSTMs, CNNs, RNNs) to classify genres and emotions in songs based on spectrogram information. Developed and fine-tuned models for high accuracy, leveraging advanced software engineering practices for model training and evaluation. Kaggle competition (Python)

Designed and built an Android app that won the Vodafone/EESTEC 'It's APP to you' 3-day hackathon. Developed a balloon-popping game with retention business logic. (Java, Android Studio)

Trained a convolutional neural network to classify between ~ 80 cats breeds. Over 60,000 cats were classified before server bills made the project too expensive to continue hosting.