• Engineered a data pipeline to collect, transform and store real-time rainfall information across Taiwan for model training using Azure Databricks and Azure CosmosDB. • Programmed an audio command recognition API in Python, which was adopted across the AIoT team for building a voice-controlled robot. • Developed an English fluency-testing system for Chung Hua University, including a CLI and data management platform.

• Machine Learning, Instructor: Prof. Hung-Yi Lee (2019, Spring) - Managed a team of 10 TA’s to design new homework and assist with coursework. - Implemented a CLI program to smoothen the process of grading programming assignments. - Established an online grading system(similar to Kaggle). The system accepts a tar file containing required files and returns a corresponding score, which is displayed on the public leaderboard and used for grading. • Machine Learning, Instructor: Prof. Pei-Yuan Wu (2018, Fall) - Assisted over 150 students with homework and questions. - Responsible for designing and grading the class final project (Automated video captioning using RNNs).

• Network and Systems(NS) Lab, Feb 2019 – Feb 2020 (1 yr 1 mo)- Research mainly focuses on multimedia networking and CDN cluster selection strategies.- Inspected network packets and behaviors using Wireshark and Google network inspection tool.• Speech Processing Lab, Sep 2017-Jul 2018 (11 mo)- Built a generative VAE model using a sequence-to-sequence architecture with GRUs and attention layers by Pytorch to transform the speaker identity of an uttered speech. - Analyzed the interpretability of latent variables in VAEs to achieve speech prosody conversion.

• Trained a speaker identification model based on a siamese architecture and Convolutional Neural Networks(CNN) using Tensorflow.• Solely designed and implemented an architecture to take the speaker identification model online, including web/API servers using Sanic, a database utilizing Redis and a model hosting server provided by Tensorflow serving.