Grant Nitta

Data Scientist @University of San Francisco

The goal of the research was to build a application that simplifies the deployment of a machine learning operationspipeline, to improve the accessibility of teaching machine learning operations in academics.• Deployed a Machine Learning Operations Python library using MLFlow, Grafana, Feast, FastAPI, Airflow, andPostgresSQL to improve ease of access in learning in academic environments, reducing setup efficiency by 70%.• Created a Cost Analysis script using Terraform that analyzes Google Cloud resource usage and calculates the fixed andpredicted variable costs that will arise from deployment, reducing the risk cost overruns by 60%.• Deployed an end-to-end MLOps cloud application in GCP, using Jinja templates for dynamic configuration, andTerraform modules for infrastructure creation, improving environment setup efficiency by 50%.