My Own Geo-based "Instagram"

These are the pictures show how the webpages look like. You can register, log in and post what you want to post.

For the images posts, the webpage can distinguish people faces.

(Frontend): a Cloud and React based Social Network

• Designed and implemented a geo-based social network web application with React JS.

• Implemented features for users to create and browse posts and support search nearby posts (Ant Design, GeoLocation API and Google Map API.)

• Improved the authentication using token based registration/login/logout flow with React Router v4 and server-side user authentication with JWT.

(Backend): a Geo-index and Image Recognition based Social Network

• Launched a scalable web service in Go to handle posts and deployed to Google Cloud (Google Kubernetes Engine) for better scaling.

• Used ElasticSearch (GCE) to provide geo-location based search functions such that users can search nearby posts within a distance (e.g. 200km)

• Utilized Google Dataflow to dump daily posts to a BigQuery table and perform offline analysis (keyword based spam detection)

• Used Google Cloud ML API and TensorFlow to train a face detection model and integrate with the Go service.

Firelook

a LBS based Android App for users to post and search nearby events

1. Developed an Android App for users to post events and search nearby events based on keyword hashtags with Java, Android Studio and Google Firebase
2. Utilized Firebase realtime database and storage to store and manage user-created content including event titles, messages, locations and images
3. Integrated Google Map API for a quick display of nearby events and navigation to user selected location
4. Integrated Google AdMob for in-app advertising to improve user experience
5. Used Firebase Cloud Function (FCF) to subscribe to new post and send notification via Firebase Cloud Messaging (FCM) to app users

Preview: Users can register an account and then login. The backend works are handled by Firebase.

Users can upload their event with an image.

Users can then review their posts.

Users can also check their locations where they uploaded the events.

Adding some comments for each event is also available.

Users can receive a push notification as soon as they upload an event with a photo.

Ticket+: Java Web Service Development-Event Search and Ticket Recommendation

• Developed an interactive web page for users to search events and purchase tickets.
• Improved personalized business recommendation based on search history and favorite records.
• Created Java servlets with RESTful APIs to handle HTTP requests and responses.
• Built relational (MySQL) and non-relational database (MongoDB), to obtain real business data from Yelp API.
• Designed algorithms (e.g., content-based recommendation) to implement business recommendation.
• Deployed server to Amazon EC2 to handle 150 queries per second tested by Apache JMeter
• Designed an interactive web page utilizing AJAX technology (HTML, CSS and JavaScript).

Features

• Authentication systems
• Liking items
• Real Time Like Updates
• Recommendation items based on user's like and location

HomePage

Website

These web pages have already been deployed on the Amazon EC2. You can check it by the link.