張泰瑋 CHANG, TAI-WEI

http://bit.do/fG82D  : [email protected] : https://github.com/david30907d : Taiwan
I'm a machine learning engineer has hands on experience of NLP, CV and bioinformatics.
  • Management Experience
    • Lead a four person team to do data labeling
  • Skill Set
    • Tensorflow, Keras, OpenCV
    • Airflow, ElasticSearch, Spark, Hadoop, Kafka, Akka
    • Postgres, MongoDB, Kubernetes
    • GCP, Azure
    • Django, React, Node.js
    • Python, Scala, Golang
    • Scrapy
    • Bioinformatics: Nextflow, Plink and basic understanding about variant calling, SNP and annotation
  • Language Skills:
    • Mandarian: native speaker
    • English: TOEIC 845
    • Japanese: N4

Selected Projects

Personalized Feed

Design the architecture of recommendation system and build it from scratch and make 10% improvement of total engagement

▪️ Using reinforcement learning 
▪️ Serve 1M people for personalized feed by using K8s and GCP Pub/Sub

▪️ Built A/B testing framework from scratch for personalized feed


Feature Store

Speed up the development process 60% by create a Feature Store service to share features between other engineering team

▪️ contributor of https://github.com/gojek/feast

▪️ Make it a robust and auto-recovery system by using Airflow

▪️ Build gRPC and RESTful as feature store serving
▪️ contributor of gojek/feast, a well-know open source feature store

Search Engine

Solve search engine precision problem 

▪️ Optimize ranking result by improving scoring function of ElasticSearch

▪️ Extract more valuable keywords into dictionary to improve Chinese text segmentation precision 

Experience

Bachelor of NCHU, UDIC Lab

▪️ Sentiment Analysis

▪️ Seq2Seq model training

▪️ GPA 3.95

Company@2x

Dcard, Junior Data Engineer, Jul 2018 ~ Mar 2020

▪️ Make 10% improvement of total engagement by designing the architecture of recommendation system and building it from scratch 

▪️ Integrate several colleague's algorithm by designing a multi-layer recommendation system and feature store from scratch
▪️ Serve 1M people for personalized feed by using K8s and GCP Pub/Sub

Company@2x

Atgenomix, Sr. Data Engineer, Apr 2020 ~ Until now

▪️ Design a general-purposed framework for Phasing and Imputation pipeline to inherit

▪️ Accelerate the best-practice NGS workflow by > 10X using Spark

▪️ Improve CI, CD process 3 times faster by using Github Action

▪️ Slim Docker image size by 30% using multi-staged build

Company@2x
Powered by CakeResumePowered by CakeResume