I am very sensitive to data and enjoy finding inspiration and ideas from them. I am proficient in machine learning, text analysis, and recommendation systems, EVM blockchain analytics, and currently use Python as my primary programming languages.
I am always open to learning new things, such as learning new data structure from blockchain. I am currently very interested in blockchain data and on-chain user segamentation. I was working in digital media, advertising (DSP, SSP, DMP platforms), gaming user analyst, blockchain data exploration like Dune dashboard.
I have many practical applications and ideas for user and customer analysis on platforms such as broadcasting networks, e-commerce, social media, CDP. For example, how to get our customers' value, how to use external data and resources to get the data more benefit, and many other practical applications and ideas. My goal is to use data to make company growth successful and get more commany user info to maximize the benefits of data.
Let Data tell the true!!!
工作經歷
Senior Data Analyst
Portto 門戶科技| Blocto • 九月 2022- 三月 2024
- Main Responsibilities:
- Establishing Data Pipeline
- Exploring new product features and competitor analysis on Dune Dashboard on the EVM
- User tagging for the Growth team (including Discord bot for monitoring
- Project details:
- Data Pipeline
- Regularly integrating client-side and BE data with external APIs and data collected by bots on Bigquery
- Establishing a systematic coding data table combined with Slack bot command manual and automatic data replenishment
- Daily data monitoring with Slack bot
- Planning client-side (app, sdk js) Amplitude event tracking to maximize data collection
- Using existing data to set company and growth team short-term indicators and conducting demos with Looker Studio for easy operation
- Segmenting and analyzing user data to maximize marketing benefits for valuable users and facilitate customer service personnel in managing VVIP customers, similar to CDP
- Using bots and APIs to collect Discord and Twitter data regularly to monitor user feedback and interaction behavior
- Dune Dashboard
- Establishing statistical standards to monitor newer or rapidly popular Dapp projects on EVM
- Regularly monitoring competitor behavior on the blockchain for specific features (ex. AA, Batch tx of Blocto (smart contract wallet))
- Collaborating with BI to plan the entire product context and ideas, explore new markets as much as possible, and establish data strategies and projects needed for growth
- User tagging for the Growth team (including Discord bot for monitoring)
- Using on-chain data to periodically observe the relationship between assets and their transaction behavior interactions
- Using data from our own app to periodically check user interaction behavior and maximize its benefits, find patterns, and find ways to replicate to bring in more users.
- Data Pipeline
Machine learning engineer supervisor
Gamania 遊戲橘子數位科技股份有限公司 • 九月 2021 - 九月 2023
- Main Responsibilities:
- Maintaining the data pipeline for game logs in Lineage M game
- Planning data projects for Lineage M
- Establishing machine learning models for Lineage M
- Designing and optimizing the company's machine learning systems and processes
- Simulating and integrating data from the game platform and headquarters for the company
- Providing reports for monitoring abnormal game behaviors for operations
- Company Awards:
- The company MVP for December 2021
- Project detail:
- Player segmentation analysis
- Responsible for detecting specific player characteristics from game player logs and using k-means algorithm to group players to achieve initial player segmentation and game quality control. Additionally, establishing a data pipeline using data mining to appropriately punish illegal behavior players, forcing them to disappear from the game and improve the game experience, monitoring abnormal game recharge behavior, and further reducing the company's property loss in the game.
- MLOPS
- Designing a complete machine learning control process flowchart and architecture, establishing a template for the initial version of the model program control with the team, facilitating data preprocessing and version control of machine learning for various projects, monitoring the quality of all data predicted by the model, further reducing engineer maintenance time and more efficiently controlling online models. Additionally, it facilitates A/B testing and online testing when developing new models, making testing to online smoother and version rollback and management more convenient.
- Tag system architecture
- Designing a player behavior tag architecture diagram, hoping to achieve player behavior control and player marketing techniques.
- Player segmentation analysis
Data Analyst Supervisor
Clickforce 域動行銷股份有限公司 • 一月 2020 - 九月 2021
- Main Responsibilities:
- Optimizing SSP ad playback quality and report production
- Optimizing DSP ad effectiveness
- Optimizing DMP report production.
- Project detail:
- Auto crawler
- Automatically plans the required machines based on the number of newly added media sites each day, and distributes machine resources to save GCP VM costs. It crawls all sites and monitors the number of crawls daily through an automated report.
- BERT Model Predicts Article Categories and Combines with Word Segmentation System, TW-IWF, and WV
- Predicts each article's category and its associated keywords and similar keyword terms for advertisement and data analysis services to provide customers with web user analysis reports.
- Automated Recommender System for CTR
- Uses a Recommender System to predict clicks based on user behavior within the broadcast network. This has resulted in a growth of 2x or more in advertising CTR.
- Predicting Gender and Age with XGBoost (Predicts User's Real Gender and Age)
- Uses cookies to predict the gender and age of a user based on their behavior on various media platforms. The accuracy rate for gender prediction is 80%.
- Auto crawler
Research Assistant
NTU Center of Genomic and Precision Medicine • 十月 2019 - 一月 2020
- Main Responsibilities:
- Paper research
- Project detail:
- RNA IMPUTATION Research
- Predixcan&S-Predixcan Research
- I wrote the draft for the paper titled "A risk prediction model of gene signatures in ovarian cancer through bagging of GA-XGBoost models"
- RNA IMPUTATION Research
Side project
Biffinex Lending Bot
- Project Goal: To use the bot to assist members in lending USD assets at the best possible rates.
- Main Responsibilities:
- Training the model with the lending rate, period, and volume.
- Current Results:
- The lending rate spread for lower-tier members is 1-3% higher than that of competitors.
- Main Responsibilities:
- Future Vision:
- Iteration of existing functions
- Implementation of new functions
- Intelligent long/short ratio robot order book.
Education
Sep 2017 - Jun 2019
National Taiwan University
Institute of Epidemiology and Preventive Medicine (majored in bio-statistic)
論文: 以機器學習方法建構卵巢癌病患之基因預後模型 (參加ICIBM國際會議)
Published on Journal of Advanced Research May 2021
A risk prediction model of gene signatures in ovarian cancer through bagging of GA-XGBoost models (quoted by 15 times until 2024-03 )
Sep 2013 - Jun 2017
National Cheng Kung University
Statistics
Data Analysis Projects During School:
- Online News Popularity
- The Relationship Between Ice Cap Area and Sea Level
- Saint Tome Malaria Mosquito
- 2018 Taipei Mayoral Election Prediction
- The Hidden Dangers Behind Terrorist Attacks
- Semiconductor Process - Wafer Image Analysis
技能
Programming Language
- Ｒ
- Python
DataBase
- Bigquery
- Athena
- Mysql
- Sql
- S3
- Storage
- Redis
Command Bot
- Slack Bot
- Discord Bot
