More than half of man-hours are occupied by AI development environment preparation and management
When working on AI development as a team, there is a lot of work involved, such as building a complex development environment, managing computational resources shared during learning, and managing artifacts such as learning models and learning results, which accounts for approximately 50% of the man-hours. is spent on environmental preparation and management. Macnica provides a service to build an AI development environment (infrastructure) so that data scientists can concentrate on model development, which is their main job, and focus on work that creates value for the company.
This kind of problem is hidden in the learning model development environment
Inefficiency due to scattered information
In the model creation phase, it is necessary to shorten and speed up the cycle of data collection, model creation, learning, and evaluation until the desired accuracy and results are achieved. However, when actually carrying out these flows, tools for data management and model creation are not linked, forcing inefficient work.
Struggling with mechanisms and development methods, not model development itself
The number of cases where AI frameworks are distributed as container images is increasing, and the use of containers in machine learning has also become common. In addition, the number of cases where container orchestration is used is increasing in anticipation of resource optimization and future resource expansion. In addition to this, managing the code of the created model, managing the data used for learning, etc. takes time for man-hours that are not the main business. and acquire knowledge.
Security risks in using containers
The use of open source frameworks and libraries in AI development is unavoidable in order to improve development efficiency, and development is being carried out using container images that include these. However, containers can harbor suspicious malware as they are collections of file packages. It is extremely dangerous to use such risky images by connecting them to your company's network.
AI learning environment building service
To address these challenges, Macnica offers proposals and construction of optimal infrastructure for AI development, as well as a UI for developers.
Features
1. Supporting AI development with a simple UI
シンプルなUIで開発に必要な操作が全てUIに纏められており、コンテナやコンテナオーケストレーションの知見がなくとも、すぐにAIモデル開発に取り組めます。
2. Functions for system administrators
It provides a mechanism to visualize server resources used by developers, manage developers, and add GPU resources in the future.
3. Minimize security risks even in a container environment
Provide an internal private registry to store container images and introduce a mechanism to share these images among team members. This prevents developers from using container images with hidden security risks.
service menu
Standard package
A plan that includes recommended hardware (specs can be selected from Shochikubai) and system construction as a package
Inquiry desk, portal site maintenance 1 year included
System construction
Portal site A plan that provides only construction, contact point for inquiries, and portal site maintenance for 1 year
full custom
A plan to propose and build a system design that meets your needs and usage
Example of service provision
For more information on what you can do with the portal site, please see this article.
price
For pricing, please contact us below.