Timeline
SOLA is a Smart Offline-first LLM Assistant. By adopting an offline-first approach to data and model execution, privacy is guaranteed. Still, its integration with Large Language Models (LLMs) and access to external (online) tools make it smart.
SOLA integrates a range of open-source speech and language models and targets the low-power Nvidia Jetson Nano as its deployment platform for inference on the edge.
SOLA - Smart Offline-first LLM Assistant
May 2024
SOLA is a Smart Offline-first LLM Assistant. By adopting an offline-first approach to data and model execution, privacy is guaranteed. Still, its integration with Large Language Models (LLMs) and access to external (online) tools make it smart.
SOLA integrates a range of open-source speech and language models and targets the low-power Nvidia Jetson Nano as its deployment platform for inference on the edge.
Impressed by how far open source computer vision models have come in the past few years, we wanted to explore their application in the field of sports analysis and coaching.
In this project, we detect the player's 3D poses and 2D positions on the tennis field. These features are then used to perform classification of the style of tennis shot, by making use of a custom architecture containing graph and recursive neural networks.
AI Tennis Shot Prediction
May 2024
Impressed by how far open source computer vision models have come in the past few years, we wanted to explore their application in the field of sports analysis and coaching.
In this project, we detect the player's 3D poses and 2D positions on the tennis field. These features are then used to perform classification of the style of tennis shot, by making use of a custom architecture containing graph and recursive neural networks.
Quantum computers hold enourmous potential for solving problems that classical computers are incapable of. However, this technology is currently held back by the instability of the quantum states in the qubit and qudit devices.
In this project, an efficient on-chip machine learning model is developed for prediction of optimal state control parameters in real-time, helping to alleviate the instability issue.
Machine Learning for Quantum Control
May 2024
Quantum computers hold enourmous potential for solving problems that classical computers are incapable of. However, this technology is currently held back by the instability of the quantum states in the qubit and qudit devices.
In this project, an efficient on-chip machine learning model is developed for prediction of optimal state control parameters in real-time, helping to alleviate the instability issue.