This service provides one-stop support for the development of machine learning products, from requirement definition to AI integration, and contributes to increasing the value of services and products to users.
Background of Service Launch
The use of machine learning to provide a better user experience is becoming increasingly important in improving the value of products.
However, in addition to the development of the machine learning model itself, successful development of machine learning products requires specialized knowledge in data science in each phase of development, such as defining product requirements that take into account the characteristics of AI and incorporating AI into the system in anticipation of model improvements. Therefore,In many cases, machine learning models have been developed, but their functionality does not fully translate into increased value for the product.It is.
Hakky's new "Hakky" product, which will be offered at this time, is called "Hakky'sMachine Learning Product Development Support ServicesIn the development of new businesses, in-house services, and products using AI,
(1) Define AI requirements
(2) Machine Learning Model Development
(3) Incorporation into products
The three steps areExperienced machine learning engineers and data scientists to assistI will do so.

Service Details
(1) Definition of AI requirements
In addition to the usual requirements definition, machine learning products require a value definition of how the AI will extend and improve the user experience, and a UI design that takes into account the characteristics of the AI, such as how to display the results output by the AI.
To accurately define such "AI requirements" requires expertise in data science, which cannot be covered by ordinary system requirements definition or UI design.
Hakky has a proven track record of servicing multiple products with AI-based functions,We excel at defining AI requirements centered on the user experience, and we are able to provide our customers with the best possible service in the field of AI.I will do so.
Machine learning model development
During the model development and validation phase, we ask ourselves "Is it really accurate? What would it take to increase accuracy?" The expertise of the data scientist is indispensable to determine "Is it really accurate?
However, each step of model development, such as data preprocessing, exploratory analysis, and feature engineering, has a "don't know until you try" nature, and accuracy cannot be improved if sufficient time cannot be spent on each step. Therefore,It may take too much time to build the model and not be able to release the product, or conversely, it may take too much time to rush into production and not achieve the expected accuracy, and this may not lead to the desired results.In many cases, this is a lost opportunity.
Hakky uses in-house developed model building tools to semi-automate time-consuming tasks, allowing data scientists to focus their knowledge on improving the accuracy of the models.Build accurate models in the shortest possible time and at the lowest possible costIt is.
(iii) Incorporation into products
Machine learning products are characterized by their ability to accumulate data to improve accuracy and user experience. However, if the model is not built on the premise of continuous improvement, it will not be able to operate successfully.
Hakky is,In the phase of productizing AI, MLOps (*) will be realized so that data scientists and engineers can work together to continuously develop, operate, and improve AI, including architecture design, infrastructure/API development, and deployment/operation flow construction with a view to improvement.We will support the following
MLOps is a management system to ensure that the three steps of model creation (MachineLearning), system development (Development), and operation (Operation) run smoothly.
Hakky has experience in launching multiple new businesses using AI, and has supported various users as a group of data science experts in contracted development of machine learning products, infrastructure construction, and operational support. Utilizing the knowledge gained from these achievements,Realizing the mission of "making products valuable with data" by accompanying users from the planning stage.The following is a summary of the results of the project.
About Hakky Co.
| name of company | Hakky Corporation |
| representative | Kazumasa Saito |
| address (e.g. of house) | Shinjuku Daikan Plaza A1018, 7-1-7 Nishi-Shinjuku, Shinjuku-ku, Tokyo |
| Business | 1. support for data utilization 2. machine learning product development support 3. support for data infrastructure construction |
| uniform resouce locator | Handbook:https://book.st-hakky.com/ Company website:https://www.about.st-hakky.com/ |








