Providing deep learning models as webservices
Description:
Setting up a webservice with a machine learning pipelines in the backend to let users/customers/project partners:
- Manage Datasets, Annotations and Models
- Execution of Machine Learning Applications, Training and Inference
- Access to high-performance compute cluster via simple user interface
- Annotation integration
Links:
None
Keywords:
machine learning pipelines, workflows, datastorage
Motivation:
Combination of software architecture, microservices and tools to use when creating ML pipelines
Requirements/Prerequisities:
docker, kubernetes, airflow, minio
Level:
concret: specific best practice (e.g. use microservice)
Application domain:
Data science (analysis & visualisation), Software engineering
Main phase:
Data Science: Preparation/Integration, Data Science: Modeling/Training/Evaluation, Development: Implementation/Code/Build, Development: Testing
Related literature:
https://airflow.apache.org;
https://kubernetes.io;
https://www.docker.com/
https://min.io
In which projects do/did you use this practice?
S3AI, FlexSpect,...
Researcher
0–2 years of experiences
Software Competence Center Hagenberg
| 1. How do you rate the potential benefit for your projects? | 5 |
| 2. How often are you using that practice? | 4 |
| 3. What is the effort to introduce the practice in your project upfront? | 3 |
| 4. What is the effort to apply the best practice in your project daily basis? | 3 |
Questions 1, 3 and 4 (1 = Low, 5 = High)
Question 2 (1 = Never, 5 = Always)