National Cheng-Chi University
Sep. 2016 ~ 2020 Senior Student
- Computer Science, GPA 3.97
- Research field: IoT Ecosystem, Edge Computing, Embedded System, Data Science (ML)
- Programming Languages: Python, C++ / C, VBA, R, SQL
- Embedded System: Raspberry Pi (Raspbian), STM32 (Mbed OS), Linked 7688 (OpenWrt)
- Others: Git, PCB design (Eagle), LPWAN (Sigfox, NB-IoT), Tensorflow
- "The 24th of the Raspberry Pi Meetup". Jan. 2019
- Attendance 100+
- "Raspberry Pi Jam". Mar. 2019
- Attendance 70+
- "Techbang Magazine PiM25 Project Sharing", Mar. 2019
- Attendance 60+
- "LASS Conference International Session", Jul. 2019
- Attendance 100+
Research Aide in Array of Things, @Chicago in United States
- The Array of Things (AoT) is an urban sensing network in Chicago city, measuring factors that impact livability in cities such as climate, air quality, pedestrian, floodwater, and noise.
- Used machine learning method to calibrate radiative error with current ambient light level.
- Built pattern recognition algorithm for time series data.
- Reprogrammed micro-controller firmware and micro-processer in the embedded system.
- Connected AoT data API and LASS data API to transfer data to each sensor network platform.
Research Assistant in Network Research Lab, @Taipei in Taiwan
- Contributed to Location Aware Sensing System (LASS), the biggest large-scale PM2.5 sensing IoT system with more than 7,000 participating devices over more than 41 countries.
- Researched edge computing and analyzed multiple time series data.
- Constructed an optimal deep learning model to forecast air quality.
- Developed PiM25 IoT device and administered PiM25 community.
- Tested the first generation of MIT air quality sensor and improved the data quality by a calibration model.
- Developed an automotive analysis platform to produce a data report of employees' personality.
- Increased efficiency of the parsing algorithm by 20%.
- Designed an employee database according to end user information and predicted employee turnover.
Seq2Seq Model Forecasting Air Quality with Edge Computing
- Constructed a real time deep learning model to predict PM2.5 data
- Reduced burden on a cloud server and improved the efficiency and accuracy of the training model
Developed a Low-Cost Devices Calibration Model of PM2.5 Measurements
- Provided a correction formula and an open calibrated API
- Utilized professional weather stations to correct low-cost sensor data
- The project proposal was accepted with College Student Research Scholarship and approved by MOST
Pattern Recognition for Time Series Data and Calibration Model Improvement (in progress)
- Applied specific K-means cluster algorithm to recognize time series pattern
- Improved calibration model to minimize radiative error caused by solar radiation
Adaptive Sample Rate in a Sensor Network for Air Quality Monitoring (in progress)
- Analyzed time series data and sought the best sampling interval to reduce energy cost
Jul. 2019 ~ Sep. 2019 System Integration
- Replicated USB-to-Serial function and uploaded the latest sketch to a self-designed PCB board
- Reprogrammed a Micro-Controller Firmware into another Micro-Processor chip
- Combined PM2.5 sensor and Raspberry Pi to develop a truly end-to-end IoT ecosystem
- Built customer data visualization, outlook design, and an efficiently OTA update mechanism
- The first Taiwan project published on the Magpi (an official Raspberry Pi magazine), and got accepted in
HKoscon (2019) and COSCUP (2019) to give a talk
- Tested and deployed more than 70+ environment sensors in Taiwan and reported the error devices
- Provided a regression model for pm2.5 sensors to calibrate data and improve data quality