See the repo here.
And the website on https://myarxiv.club.
This is a full stack web app that utilized various technology stacks, including docker, docker-compose, nginx, gunicorn, supervisor, mysql, redis, celery, flask, sqlalchemy, alembic, pytest, vue.js, uikit3, webpack4 etc. For the philosophy on the architecture of this web app, see this post.
The backend engine employed various NLP techniques such as rake and fuzzy matching. For more NLP attempts on arXiv data, see this post.
The website is deployed via cloudflare CDN with nginx alone at VPS in the US, while the app is actually running on the machine back in China.