The Singapore Institute of Technology (SIT) has announced the establishment of its Centre for AI to undertake research and development (R&D) projects for enhancing AI adoption across industries and government and help in tripling the number of AI practitioners within the next five years.
This will equip corporations, including SMEs, with the tools and support required for successful AI integration.
It is currently operating in interim facilities while its permanent space in SIT's Punggol Campus is under construction.
SIT has built the centre in collaboration with Nvidia.
The centre will provide technological support, innovative solutions, and co-supervision by SIT and NVIDIA experts to industry partners.
It aims to assist the industry with the customisation and refinement of foundational and generative AI models and develop AI applications to meet specific needs.
According to SIT, the centre will present opportunities for industry partners to testbed AI products, and support project collaboration on NVIDIA AI software solutions.
This includes NVIDIA AI Enterprise software including NVIDIA NeMo, a platform for developing custom generative AI models, and includes tools for training, retrieval-augmented generation, guardrailing and toolkits, data curation tools, and pretrained models, it added.
Industry projects
SIT has initiated over 10 industry projects in domains including protein discovery, computer vision, and safety enhancement in transportation.
It is working with SMRT Corporation, a multi-modal public transport operator in Singapore to develop a generative AI (GenAI) system for enhancing safety investigations and reporting.
By leveraging GenAI, the incident reporting process can be automated, expediting the documentation process, and acts as an additional layer of check that enhances the completeness and accuracy of incident reports.
It benefits ground staff involved in incident reporting and investigation and helps facilitate analysing the incidents, raising warnings, and extracting actionable intelligence.
The institute is also collaborating with CTMetrix to develop GenAI solutions for creating training data for multi-view X-ray scans of faulty electronic printed circuit boards (PCBs).
This data trains an image inspection AI to accurately detect defects in soldered connections, overcoming challenges like out-of-focus alignment.
It facilitates adaptation to new PCB designs without requiring full model retraining.
