AI Solution

The vendor is required to provide to deliver a solution, consisting of software, AI models and engineering services, deployed on the province’s artificial intelligence (AI) server infrastructure.
- Solution will leverage the existing compute AI server infrastructure consisting of a cluster of 32 NVIDIA h100s GPUs with an aggregate capacity:
• Total GPU memory: 2,560 GB (2.5 tb) hbm3,
• Total memory bandwidth: 107.2 tb/s,
• Peak fp16 compute (no sparsity): 63,328 TFLOPS (63.3 PFLOPS),
• Peak fp8 compute (sparse): 126,000+ TFLOPS (126 PFLOPS), and
• Inter-GPU bandwidth (NVLINK): 28.8 tb/s total potential (assuming full mesh/topology support), that will support the development, deployment and continuous operation of multiple large language models (LLMS), including open source and closed sourced models
- Requitement:
• Enabling ai-assisted processing of sensitive data in accordance with the province's data security classification standards and in compliance with the country privacy laws, through secure processing of datasets with high sensitivity and robust systems monitoring for abuse or misuse. 
• Supporting a wide range of use cases, including document summarization, fraud detection, and archival data analysis, through selection, configuration and fine-tuning of both closed and opensource models, with ongoing support for loading new models as technologies evolve. 
• Providing both interactive (chat-based) and automated (agent-based) modalities of AI service delivery, with configured connectivity protocols and performance monitoring systems. 
• Leveraging province-owned hardware, including NVIDIA h100-based compute infrastructure, to host large language models (LLMS) and other multimodal models through deployment to infrastructure and configuration of connectivity protocols. 
• Scalability of the proposed solution, if required, due to future growth of AI server infrastructure, supported by deployment architecture and performance monitoring for capacity planning. 
• Enabling long-term data sovereignty and compliance with post-quantum encryption standards through secure processing protocols and robust systems monitoring. 
• Implementing and reporting on clearly defined metrics for privacy assurance and system performance, such as data breach incident frequency, AI workflow completion times, and audit trail completeness through comprehensive monitoring and evaluation systems. 
• Establishing documented risk mitigation framework and contingency strategies for AI server infrastructure services, including response and recovery protocols for service failures or security breaches through secure processing protocols and comprehensive monitoring systems.
- Contract Period/Term: 3 years
- Questions/Inquires Deadline: July 11, 2025