Safir Zeka (Safir Intelligence) is a Machine Learning as a Service product that is one of the actively developed and maintained projects within B3LAB, which provides web based distinct services for specific use cases. Safir Zeka serves the most basic libraries and customized out-of-box models for software developers. It is also optimized for data scientists and end-users to help them develop Machine Learning applications. It uses the power of container-based virtual environments to host both development environment, called notebook. Through the notebook interface and safir-pipeline, software developers and data scientists are able to create their own customized procedures and install any new software in their own.
Safir Zeka provides the following pipeline:
1) Providing data from external sources using third party APIs (S3, Twitter, Facebook etc.).
2) Uploading data from local sources.
3) Data preprocessing and normalization.
4) Data visualization tools.
5) Creation of machine learning models, training and testing of these models
6) Easy deployment of developed models.
It is also possible to automatically import models produced by the process wizard. Safir Zeka also has out-of-box machine learning models for end-users. These models can be accessed via their APIs.
Within the scope of the Safir Zeka project, research and development activities are continuing within B3LAB and it is aimed to maximize user experience by adding new features.
Through the Safir Zeka platform; the models, which are proven by academic studies and are compiled from high-quality publications, are presented on the image, video, text and flowing data and presented to the end-user in the form of web services.
In addition to the general solutions for the end user, consultancy services are provided by producing on-site detection and customized models for solutions to corporate needs. It is possible to perform analysis and forecasting by sending the models to the data center in order to process the data that will be a problem with privacy or cloud loading.