Institute of Smart Systems and Artificial Intelligence (ISSAI) at Nazarbayev University has developed the Mangitas series. It is a comprehensive, secure AI solution designed to operate entirely within air-gapped or government-certified environments. Working within resource constraints while leveraging strong internal expertise, the Mangitas Inference Server was assembled from commercial-off-the-shelf components in order to deploy AI models such as Oylan, Mangisoz, Beynele, and Tilsync , it now powers these models in production.
Mangitas is more than just a high-performance server, it is a key to unlocking AI’s potential without compromising on security. Its core innovation lies in packaging large-scale, complex AI models into a compact, reliable, and secure hardware form.
The Mangitas series by ISSAI is a custom-built, secure AI inference server engineered for government and enterprise sectors where data sovereignty and cybersecurity are non-negotiable. The security dilemma of modern AI lies in the fact that, while artificial intelligence has become essential for digital transformation, it simultaneously presents a critical challenge for public institutions and enterprises operating in sensitive sectors. The demand for advanced AI capabilities often collides with the absolute requirement for data security and national sovereignty. At the same time, cloud-based deployment is frequently prohibited due to the risks associated with sensitive data exposure, national security concerns, and strict regulatory compliance.
ISSAI offers two distinct models configured for different operational needs. Both are built on the same foundation of security and performance.
The Mangitas 01 is configured for organizations that require a powerful, dedicated solution for a single, mission-critical AI function. It provides a focused and highly optimized environment for specialized applications.
Best For: Secure, single-purpose AI operations.
Supported Model: Mangisoz, our powerful speech and translation suite for real-time transcription and translation between Kazakh, Russian, English, Turkish, and Chinese.
Core Use Case: Performing secure, real-time, and local language processing tasks without exposing sensitive data to the internet.
The Mangitas 02 is a versatile AI powerhouse, configured to run the full suite of ISSAI’s advanced AI models simultaneously. It is the ideal solution for organizations aiming to establish a comprehensive, multi-purpose AI hub.
Best For: A full-spectrum, multi-purpose AI hub.
Supported Models:
Core Use Case: Deploying a complete, sovereign AI ecosystem for diverse and complex tasks, from data analysis to content creation.
Every Mangitas server is built on four pillars of trust and performance:
Chassis: Custom-modified 5U rackmount enclosure adapted for airflow and multi-GPU layout.
AI Acceleration: Features a multi-GPU system (4x configuration) with 192 GB total VRAM.
Compute Core: A high-performance, multi-core workstation-grade processor with a liquid-cooled solution.
Memory: 128 GB of large-capacity ECC memory with a high-bandwidth DDR5 architecture.
Storage: 8 TB enterprise-grade solid-state drive with an NVMe interface for high-speed loading.
Power: Dual-redundant 3.3 kW high-wattage PSU configuration with smart power management.
Proven and Trusted by National Leaders.
The Mangitas platform is not just a concept, it’s a field-tested solution. It is already deployed for internal use in educational and quasi-governmental institutions.
Contact our team of experts to discuss specific requirements and learn how the Mangitas server can power your digital transformation securely.
issai@nu.edu.kz
