Hamburger Icon
GPU Management
Solutions for AI
Machine Learning
A Kubernetes-based GPU infrastructure
management solution that ensures
machine learning workloads operate
at maximum efficiency.
AstraGo maximizes GPU server utilization
and improves infrastructure management
with a job scheduler, resource optimization
technology, real-time monitoring, and
It stands out as the sole solution
offering predictive learning time for added
1 of 2
Main Features
Machine Learning
Support System
Astrago supports learning in three ways:
Model Hub, Built-in Image, and Custom,
to match the user’s level. Additionally, it
contributes to the time management of
researchers and projects with unique
ML time prediction feature
Model Hub
SOTA models can be acquired by changing
parameters in a GUI setting, which is perfect
for educating students and beginners.
Built-in Image
Decrease the time required for setting up
the development environment by providing
images that include framework libraries
optimized for GPU.
Users have the ability to upload a picture of
the frameworks and libraries they prefer and
utilize them.
ML Time Prediction
Provide an estimated completion time when
uploading a model, and see real-time progress
in a bar-like UI during training.
and Orchestration
Improve productivity throughout your
infrastructure by utilizing job schedulers
and optimizing resources, while also
gaining valuable information on activities
through monitoring and reporting.
Resource Optimization
Validate and reclaim underutilized workloads
to prevent resource waste
Provide insightful reports based on
analyzed statistics and data to diagnose
the current situation
Job Scheduler
Automatically distribute jobs to optimal
GPU resources
Provides resource usage by cluster/node/ workspace, including GPU, CPU memory, etc
vidigo-h-0vidigo-h-1vidigo-h-2vidigo-h-3 procon-img
Key Features
& Differentiators
Key Features. I
Job scheduler that supports
infrastructure operations tailored to
user-defined objectives
An efficiency-focused job scheduler enhances
GPU utilization by prioritizing nodes with
lower available resources when assigning jobs.
Assistance for different GPU partitioning
technologies, such as MIG, enhances
the efficiency of scheduler operations.
Key Features. II
Differentiators. III
Differentiators. IV