User:Taylor Deng

Taylor Deng

Early life
Born in Hubei, China....

2012.9-present Samsung Electronics Senior Engineer

 * Context-aware Modeling: Built context-aware execute engine using domain modeling engineering.
 * Device-Centric Cloud: Built a mobile cloud for personalized data processing and sharing across devices, and providing personalized service such as Advertise.
 * Innovations: 2 patents proposed and 1 paper published.
 * Contributions: DSL in Context-aware project.
 * Skill Set: DSL, Context-aware framework; Personalized Service; Device Centric Cloud.

2009.6-2012.9 Cisco Search Engine Engineer

 * HA/HC/Monitor TO: Primary/GSB policy for HA, Float IP for service redundancy. SNMP module for cluster monitor. *Multi-Shard policy for system scaling out.
 * NRT (Near Real-Time) TO (Task Owner): 700+ QPS per node, 3s delay after CRU/D can be searched. Viterbi Algorithm was used to merge index file. Support cross DC synchronized index file and search.
 * Search Facet TO: Can handle 100M records per node do faceting. Support single term/intersection/path faceting. Used bit-set to filter, and recursion approach in design speeds up computing, only 4ms performance consumed compared without it.
 * Analyzer Features TO: Design analyzer based on Lucene, including segmentation and tokenization for input/output according Lucene versions, features including Stem, Synonym, Multi-Numeric Range query, Payload etc.
 * Online DB Service (Designer): Designed logical schema mapping with Colum Family. Implemented gateway to handle request operating HBase for different client. Use coprocessor components to deliver message and bulk load to import data.
 * Innovations: 1.) Bernoulli Experiment was used to optimize Select Top-N from multi-node. 2.) Viterbi Algorithm was used to optimize Near-Real time search.
 * Contributions: 1.) 2T/day log indexing search solution. 2.) System troubleshooting and diagnosis. 3.) Dynamic programming algorithm used in merging index file. 4.) Lucene feature design and optimize.
 * Skill Set: Distributed System Design and Architecture; Search Engine. HDSF/ HBase/ Zookeeper; System Profile and diagnosis; Scaling out; NoSQL DB; Design Pattern; MQ; Algorithm;

2007.6-2009.6 Tudou.com Search Engine Engineer

 * Chinese Segmentation/Search Suggestion/Search Relevance: 1.) To 1/5 in shorten index time and 10 times faster in segmentation. 2.) Search UE enhancement including characters conversion, traditional/simple Chinese conversion, and date/ number translation, and Pinyin error correction. 3.) A nested prefix map was designed to achieve Search Suggestion.
 * Video Classification/ User Score System/ Video Channels: 1.) Implemented Refined Centro Classifier Algorithm to classify videos into deferent channels. 2.) Analytic Hierarchy Process was used as mathematical modeling to establish the User Score System.
 * Awarded: 1.) Best Practice Award-2008.4: Search Project guided by project refactor and Efficient JAVA, this optimize reduced at least half (at least 6) servers on production. 2.) Best Performance Award-2008.6: Search Suggestion: 2 servers for 6000+ QPS on production.

Education Background

 * 2005.9-2007.12 Master of Software Engineering, University of Science and Technology of China.
 * 2001.9-2005.7 Bachelor of Information Technology, Yangtze University.