Zejun Lin

* Implement basic functions of Git supervised by Bjarne, including an Input Parser dispatching different commands, a family of Git Objects using OOD that keeps track of (staged) files, commits’ tree, branches & tags with SHA-1 for hashing.

* Lead several AWS group projects including the Dining-Concierge-Chatbot using S3, API Gateway with swagger, Lambda, DynamoDB to build a web application, Lex to build NLP model, and Elasticsearch for fast indexing ---- (https://github.com/Crysple/Dining-Concierge-Chatbot)
* The Smart door project uses the above techniques with KVS/KDS and Rekognition to build a face authentication system ---- (https://github.com/Crysple/smart_door)
* The Photo Album project supports photo queries using… Show more

* Lead several AWS group projects including the Dining-Concierge-Chatbot using S3, API Gateway with swagger, Lambda, DynamoDB to build a web application, Lex to build NLP model, and Elasticsearch for fast indexing ---- (https://github.com/Crysple/Dining-Concierge-Chatbot)
* The Smart door project uses the above techniques with KVS/KDS and Rekognition to build a face authentication system ---- (https://github.com/Crysple/smart_door)
* The Photo Album project supports photo queries using natural language through both text and voice. Besides we also build Code Pipeline connected to Github and write Cloud Formation template for continuous delivery ---- (https://github.com/Crysple/photo-album) Show less

* N-Puzzle Board Game: Implemented and compared breadth-first, depth-first, A-star search algorithms
* Sudoku: Implemented backtracking search with MRV(minimum remaining value) heuristic to reduce variables domains
* 2048-Puzzle Game: Implemented an "Adversarial Search" agent to play using "Expectiminimax" and "Alpha-beta pruning"

* Lead a group of 6 to deployed HDFS (Hadoop Distributed File System) with HA (high availability) on dozens of machines (including Zookeeper and enabled Resource management, Journal Node, and a standby node for hot-swapping)
* Generate up to five billion records, respectively deploy Apache Spark, Kafka & Flume to simulate users’ real-time login/logout behaviors and compute statistics like pv, uv using MapReduce, and finally, show real-time output on a designed webpage.

* Lead a team of 3 to participate in this competition, which aims to predict customers’ current location (in terms of which shop they are in within a mall) according to information like the wifi list, the latitude & longitude on their phone.
* After data cleaning & mining, develop multiple models integrated with Stacking (rank 76th/2845 in the preliminary round). Then utilize SQL, MapReduce, UDF on Ali’s Platform to reimplement our ML models on tens of TB data (rank 14th/100 in the final).

* Independently employ MVC framework on XAMPP to develop a website for expert recruiting. Use PHP as backend language & apply bootstrap to decorate fore-end. Design a complex database from Entity model, optimize & implement it using MySQL.