auenkr / email-classifer

A email (only gmail) classifier and summarize using LLM

Gmail Classifier with LLM

1. Multi-class Classification:

• Classify emails into categories like Important, Promotions, Social, Marketing, and Spam based on LLM insights and keywords.

2. General Fallback:

• If the LLM analysis is inconclusive, categorize the email as "General."

To start Project locally

Pre-requisites:

1. Node or Docker
2. Google OAuth account setup with scope more detail

     - openid
     - email
     - profile
     - https://www.googleapis.com/auth/gmail.readonly
     - https://www.googleapis.com/auth/gmail.addons.current.message.readonly
   

3. OpenAi API key

Steps

copy .env.sample to .env and get all environmental variable

Using Docker

For Linux user run docker compose up

For Windows/Mac docker-compose up

Mannual Install

Change postgres connection string also

npm install
npm run db:generate && npm run db:migrate
npm run dev

email-classifer's Issues

Sort by diff labels

Highlights diff type of label

Optimise the email fetch query

1. Without saving on db
2. Cache the old result
   Eg: if initial latest 10 email render then fetch query should no ho for these again
3. Implementation it in both backend and frontend

Create a homepage

CiCd pipeline

1. Lint
2. Docker image push to docker hub
3. Pull image into ec2 mc and run it

Add different label description by user

User address diff type of label by giving description and llm classify into them

Use Gemini api