Self-Hosted LLM Sentiment Analysis Platform on Azure Kubernetes Service (AKS)
The goal is to build a complete end-to-end solution with:
- A Flask Sentiment API
- A clean LLM Adapter layer
- Ollama (phi3:mini) running inside the AKS cluster
- Terraform for IaC
- Traefik Ingress via Helm
- Multi-stage Azure DevOps YAML pipeline
I will first test this by running scripts in the Whizlabs sandbox, after that I will automate this via an Azure DevOps pipeline.
This is a great opportunity to combine my daily DevOps work with knowledge from AZ-104, AZ-204 and AI-102, moving from an external Microsoft endpoint to a fully self-hosted AI workload.
I'll share the progress and key learnings as I go.