RESTON, VA, July 15, 2025 – MetroStar has developed a new approach to deploying artificial intelligence (AI) in secure, low-connectivity environments used by the U.S. government. Existing systems rely on cloud access, high-end hardware, or lack necessary security credentials. This solution addresses those constraints, offering a deployable model that fits operational and compliance requirements in air-gapped and mission-critical settings.
MetroStar’s Innovation Lab addresses these challenges with a hybrid architecture featuring NVIDIA H100 GPUs and RTX A5000 GPUs, enabling secure model training, fine-tuning, and real-time inference in air-gapped environments.
“We’re not just experimenting. We’re executing,” said Ali Reza Manouchehri, CEO and Co-Founder of MetroStar. “Our Innovation Lab delivers AI at the edge in air-gapped networks and across the most secure government environments. By combining H100s for secure fine-tuning with A5000s for efficient inference, we’ve built a repeatable AI deployment model that helps agencies respond to evolving requirements – faster, smarter, and more securely.”
MetroStar’s Edge Advantage: NVIDIA H100 Training + NVIDIA RTX A5000 Inference
In a recent deployment of a Retrieval-Augmented Generation (RAG) solution, MetroStar’s Innovation Lab leveraged NVIDIA H100 GPUs to fine-tune AI models on sensitive datasets in a secure enclave. Once optimized, these models were deployed on RTX A5000 GPUs in an air-gapped, low-power environment with NVIDIA NVLink, enabling real-time inference without cloud connectivity.
NVLink is NVIDIA’s interconnect technology for high-bandwidth, low-latency GPU-to-GPU communication. In systems with NVIDIA RTX A5000 GPUs, NVLink allows multiple GPUs to share memory and transfer data, functioning as a unified resource. This architecture supports flexible computing, portability, and performance suited to specific tasks.
Transforming Secure Government Infrastructure with Hybrid GPU Architecture
MetroStar uses a hybrid GPU system to support centralized, cloud-based AI infrastructure with an air-gapped design for government and infrastructure applications. NVIDIA H100 GPUs provide computing for training and refining LLMs and computer vision models within security enclaves. NVIDIA RTX A5000 GPUs, enhanced with NVLink, enable high-throughput, parallel inference for secure edge infrastructure where power, size, and security requirements are key factors. This setup establishes a modular framework to support AI tasks in disconnected or restricted environments.
Secure and Scalable AI for Sensitive Government Operations
MetroStar’s RAG solution – deployed across American-controlled networks – exemplifies secure, modular AI in action. Built on open-source frameworks and accelerated by NVIDIA accelerated computing, the solution ensures Controlled Unclassified Information (CUI+) remains within its enclave while enabling real-time reasoning.
This approach reduces time to look up information manually, cross-check data, and validate forms, making these tasks faster and easier. MetroStar’s solution streamlines complex processes, so agencies can focus on initiatives rather than tasks.
“We are architecting AI solutions that align with the modern needs of secure government environments,” said Vy Truong, CINO at MetroStar. “By integrating NVIDIA accelerated computing across secure training and edge inference workflows, we give agencies the tools to prototype, deploy, and scale AI wherever their missions require—whether in a secure facility, cloud enclave, or other sensitive locations.”
A Blueprint for AI at the Edge
MetroStar’s Innovation Lab operates as an AI platform built for secure environments. Its hybrid setup uses both NVIDIA H100 and RTX A5000 GPUs to support deployment in government and regulated industries.
By using off-the-shelf, export-compliant hardware, MetroStar provides AI capabilities with a reduced infrastructure footprint compared to traditional models.
Source: MetroStar
About MetroStar
MetroStar, founded in 1999 and headquartered in Reston, VA, is a U.S.-based AI-enabled services company focused on the public sector. The company provides solutions in AI, cloud and enterprise IT, digital experience, and application modernization to public-sector clients – from defense and national security to federal civilian, financial services, and health IT agencies. MetroStar delivers data-centric solutions, including DevSecOps platforms, to improve mission outcomes for agencies like the U.S. Treasury, IRS, USDA, DoD branches, USCIS, DOJ, FBI, Marine Corps, and NIH. The company operates an internal innovation lab and develops tools for legacy system migration and digital transformation. As of 2025, MetroStar employs approximately 500 to 600 people and has estimated annual revenue of over $100M. Its recent initiatives include AI-driven development platforms and technical support for federal IT programs.
About NVIDIA
NVIDIA Corporation, based in Santa Clara, CA, is a U.S. technology company specializing in the design and production of graphics processing units (GPUs). Its hardware and software solutions support a range of applications and simulation. Operating for over 30 years, NVIDIA has seen strong financial growth, reporting $39.3 billion in revenue and $22.1 billion in net income for the fiscal quarter ending January 2025. Its headquarters are designed to promote a flat organizational structure that encourages open communication and collaboration between leadership and staff across industries. In gaming, its GPUs power high-performance visual rendering. In artificial intelligence and high-performance computing, NVIDIA provides the infrastructure needed for training and deploying large-scale models. The company also contributes to the automotive sector with systems for autonomous driving and supports robotics with tools for AI-based perception.