Rafael Muñoz González

The following tasks were performed:

1. Architecture Design for creating a framework to streamline and standardize the use of classical machine learning.

2. Creation of a library containing the entire generic flow used in model training, validation, comparison, and deployment. The library contains classes with abstract functions for developers to complete these customizable parts when importing the library, allowing them to adapt their case to the normal flow.

3. Creation of a library containing the generic flow used to consume the model. The library contains classes with abstract functions for developers to complete these customizable parts when importing the library, allowing them to adapt their case to the normal flow.

4. DevOps implementation of the libraries using continuous integration and deployment.

5. Creation of a template repository that uses the library to train, validate, compare, and deploy the model, establishing classes to be implemented by the developer that modify the abstract customizable functions.

6. Creation of a template repository that uses the library to consume the model, establishing classes to be implemented by the developer that modify the abstract customizable functions.

7. Creation of a template repository to create an API that uses the library to consume the model in real-time, establishing classes to be implemented by the developer that modify the abstract customizable functions.

8. Implementation of generic CICD shared by all projects that will use the framework.

I participated in the Analytics Center Platform team, responsible for improving and generalizing methods for carrying out data and artificial intelligence projects.

The following tasks were performed:

- Improved developer workflows by standardizing CICD processes and automating tasks such as API deployments to API Manager.

- Provided support to developers with incidents and improved team development by indicating best practices and advice on architecture and development.

- Created a resource deployment pipeline using Bicep and PowerShell with Azure DevOps.

- Provided templates for different types of developments.

- Resolved permission issues.

- Improved the development of function apps.

- Improved the development of web apps.

- Improved the development of data engineering processes in Databricks and Data Factory.

The following tasks were performed:

1. Design and implementation of serverless architecture for executing intensive workloads external to a web app, proposing the use of containers in Azure Container Instance, Azure Batch, and Azure Functions.

2. Implementation of a message broker with a message queue pattern in Azure Service Bus.

3. Implementation of functionalities in Flask API in Python for extracting and processing geographical data in parallel using Azure Batch with an orchestrator-worker architecture and implementing auto-scaling.

4. Creation of tests for the API.

5. Creation of a self-hosted Azure DevOps agent to perform tests in a private virtual network and access protected resources using Docker, Azure Container Instance, and Azure Pipelines with CICD.

6. Modification of CICD pipelines to implement new functionalities.

7. Addition of test execution and code coverage publication to CICD pipelines.

8. Implementation of a dasymetric interpolation of statistics on geographical data using Geopandas and Tobler.

The technologies used in this project were Azure Container Instances, Azure Kubernetes, Azure Functions, Azure Service Bus, Flask, Python, Geopandas, Pandas, Azure DevOps.