About

Connor Hsu

Curious about data and real world, build product to solve problem, make machine learning into product, writing is my interest.

Experience

Senior Engineer II, Appier, Jan 2019 - Present

Data governance team / agile coach: Coordinating cross-team product features, burning down company level technical debts.

AI backend: Building extensive machine learning / analytical services through various approaches with F2E, Data Scientists, Data Infrastructure team.

Software Engineer, Appier, Jun 2014 - 2018

Build RTB bidding algorithm in a fast-growing, dynamic business environment.

Conduct experiments on real product to make daily improvements and achieve business goals.

Solve critical issues immediately and root cause analysis in uncontrolled, unreplayable, unbalanced data environments.

Enable product features with petabytes level data.

Experienced in data tools and operations (Spark, AWS RDS, Airflow)

Research Assistant, CSIE, NTU, Oct 2012 - May 2014

Build an online video retrieval application, including retrieval algorithms and backend systems.

Projects Detail

Ad Product Technical Debt Burn Down '18Q4

Co-work with scientists to migrate an legacy machine learning project from python2 to python3
Design, burn down and implement new log patch framework to secure safe patch behavior, also enable abstraction on log patch mechanism.

Data Governance '18Q2 ~

Form a Data Governance committee with tech leads to consistently improve data quality and availability. Focus involves: schema change, legacy deprecation, data platform migration planning, ..., etc.

Automatical Refund System '17Q2 ~ '18Q2

Saved more than 10 millions TWD dollars for our business as well as tremendous human effort, milestones include:

Automate process to save support team and CM team's human effort (17'Q2)
Eliminate major data discrepancy (17'Q3)
Support various timezones, formats and make debug efficient. (17'Q4)
Different dimension breakdown and co-work with F2E to build a new UI (18'Q1)

Projects Detail (cont'd)

Pipeline Reconstruction and Migration '17Q2 ~ '17Q3

Reconstruct ad-hoc pipeline and improve it by applying unit test, migrating DB, code-refactoring, and migrating to Jenkins.
Co-work with team members to migrate critical production pipelines to Airflow, till 2019Q1, more than 20 data pipelines are operated by Airflow.

Improve ML Model Performance '16Q2

Improve high quality inventory prediction by adapting suitable ML model: precision achieve 79% from 5.3%, volume increased to 12.8x
Extend CPA model to different scenario: CPA reduced to 68%, volume increased to 2.4x

Production Skills

Data Pipeline / ETL Tools: Python, Spark, Scala, MySQL (AWS RDS), AWS S3, Jenkins, Airflow, Grafana

API Services: Flask, Swagger (Flasgger)

CI/CD: Pyflakes, Pylint, Vulture, Coverage, Travis CI, Coveralls

Machine Learning: TensorFlow Serving

Other Skills

Side Projects

Bots: Slack bot, Line bot, Twitch bot

Web & Visualization: d3.js, bokeh, jQuery

Writings

Education

National Taiwan University, Taiwan, Sep 2009 - Jun 2011

M.S., Department of Computer Science and Information Engineering

National Chiao Tung University, Taiwan, Sep 2005 - Jun 2009

B.S., Department of Computer Science