An internet-based news aggregator, providing hot news scraping on popular news sources, with recommendation feature based on users' preference with the help of Machine Learning.
-
Front end:( React, Node.js, JWT)
- Built a responsive single-page web application for users to browse news (React, Node.js, RPC, SOA, JWT)
-
Back end:(Python RPC, MongoDB, Redis, RabbitMQ, TF-IDF, NLP)
- Service Oriented, multiple backends serving via JSON RPC
- Implemented a data pipeline which monitors, scrapes and deduplicates news
-
Machine Learning back end: (Tensorflow, DNN, NLP)
- Designed and built an offline training pipeline for news topic modeling
- Implemented a click event log processor which collects users' click logs, updated a news model for each user
- Deployed an online classifying service for news topic modeling using trained model
