EmotiAI

Emotion classification based on conversational data

Overall architecture

Prerequisite

• Install these packages:

brew install buildpacks/tap/pack
brew install postgresql@17
brew install minio/stable/mc
brew install helm

• Create a new python environment using conda with python version >= 3.10:

conda create -n emotiai python=3.10
conda activate emotiai

• Install required dependencies:

pip install -r requirements.txt
pip install -r serving/requirements.txt
pip install -r kserve/requirements.txt
pip install -e .

Installation and setup

Development

• To run everything locally on Docker desktop for the first time, just run: make up.
• Without build: make up-without-build.
• Down everthing: make down.
• Access Airflow UI:
• Access MLFlow UI:
• Access Minio UI:
• Access Prometheus UI:
• Access Grafana UI:
• Access Jaeger UI:

Cloud deployment

Setup Jenkins CI/CD on GCE using Ansible

• Run this command make deploy_jenkins to deploy Jenkins CI/CD on GKE.

Setup others services on K8S using Terraform

1. Run this command make deploy_k8s
2. Go to minio service and create a bucket to store preprocessed data

kubectl exec -it "REPLACE WITH MINIO POD NAME"  -- /bin/bash
mc alias set myminio "MINIO URL" "MINIO ACCESS KEY" "MINIO SECRET KEY"
mc mb myminio/emotiai

3. Forward-port PostgresQL and run all scripts in folder /data