1.2 Micro services
- KDD –Knowledge Discovery in Databases is the organized procedure of recognizing valid, useful, and understandable patterns from large and complex data sets.
- DataPrep –DataPrep offers various transformations/operations for a variety of data types – tabular, image, audio, video and text. Every data type has its own set of transforms/augmentation/visualization capabilities currently supporting 125+ such functionalities.
- Data Annotation –Data Annotation helps in labeling and assigning the ground truth to the data available in various formats like text, image and video.
- Generate AI Model –Train AI models using Transfer Learning or QpiAI Base Models suited for your domain and sub-domain with a few clicks. Also, create your own Custom Models for unique tasks and share it with collaborators working in a particular domain using Tag and Submit.
- Model Discovery – Discover new neural network architectures and create novel models tailored to your domain and dataset. Create models with multiple design objectives in mind for example accuracy, inference latency and model size.
- Model Predict –Use trained models to make predictions on custom datasets and validate its performance.
- Deploy AI Model –Model Deployment microservice is used to package QpiAI generated model objects so that it can be deployed on a cloud/edge device effectively.
- Monitor AI Model –Use model monitoring service to monitor the generated models and see how the dataset shift (if any) has impacted the model.
- Edge Device Manager – Connect edge devices to the QpiAI Pro platform over the air and deploy/manage/terminate containers on these devices with the click of a button. The devices are monitored periodically for their CPU and RAM utilization with statistics displayed on the UI.
- View History –All the above mentioned microservices requests can be monitored through a View History service.
- Tag and Submit –Custom Models can be tagged to a specific domain and subdomain and submitted to organizational git repo through Tag and Submit.
- Tag –Tag microservice just lets you tag QpiAI basemodels to a different domain.
Before initializing any of these microservice, our base scheduler ensures that these requests are fired only on the free VMs which are not utilized.
In this User Guide, each microservice is explained in detail and sample datasets/files are provided in the Appendix section.