University of California, Irvine (UCI)
Master of Computer Science
Ting-Wei Lin
University of California, Irvine
Ting-Wei Lin
I am Jacky from Taiwan, currently a Master's Student in Computer Science at the University of California, Irvine.
I am always into the community side of Computer Science, also have equipped with insights derived from fields beyond software development. From my internships at Amazon and DHL Express and research experience in Academia Sinica, I am interested in utilizing big data techniques in systems and integrate them into people’s lives.
I am now looking for Full time Software Engineer opportunities in the spring of 2022. If your organization is ambitious and indeed want to make the world a better place, USE ME. Let me help! :)
-
Jun 2021 ~ Sept 2021Amazon Web Services, Amazon.com
Software Development Engineer Intern
- Designed Points of presence (POP) capacity measurement on the basic of requests per second and network bandwidth in bytes per second, improving 10% accuracy on projecting big events’ CPU utilization
- Integrated POP’s capacity into a full-stack system with heatmap layout with React.js, Express.js, and AWS DJS, RDS, RedShift, S3, API Gateway, and Lambda, accurately displaying status of each POP and host
- Promoted POP capacity measurement system, increasing 30% efficiency of related teams’ works for checking POP’s status
-
Oct 2019 ~ Apr 2020Computational Linguistics & Information Processing Lab., Academia Sinica
Research Assistant
- Built deep neural network architecture—XRR—for risk ranking with financial reports, yielding up to 7.4% improvement according to two ranking correlation metrics and published at AAAI-20 Workshop on KDF
- Created platform for investors or financial experts to spot risk-related information and facilitate analyses of soft information in financial reports by deploying Express Web framework
- Extracted and analyzed crucial factors (User behaviors and User preferences) on a real-world TV dataset and formally defined the problem of TV Recommendation
- Proposed algorithm to aggregate nearby item content embedding, time series, and channel preferences, yielding up to 20% improvement in recommendation quality in linear TV services
-
July 2017 ~ Jan 2018DHL Express
Software Engineer Intern
- Developed system with JavaFX to compare and calculate staff salaries automatically and generate reports, increasing efficiency of Department of Human Resources by 50%
- Engineered E-Government Uniform Invoice (EGUI) System for Department of Finance with Struts 2 Web framework, replacing old invoice system in response to new government policies
- Trained staff on functions and operations of EGUI system and installation of point-of-sale machine at 15 branch offices
-
Jan 2017 ~ Jan 2018Mobile Computing & Network Communications Lab., NCCU
Undergraduate Research Assistant
- Researched wireless computer networks, encryption algorithm, and Bluetooth Low Energy (BLE)
- Proposed wireless network protocol using 3-way authentication, 2nd-party inspection, and RSA algorithm, ensuring complete data packet transport and avoiding double spending
- Created BLE mobile payment system, Beepay, including mobile application for P2P wireless transactions and back-end system website
-
Jan 2017 ~ Feb 2017Giant Shine Information Co, Ltd.
Back-End Developer Intern- Combined administrative system with I-Report to generate official school documents automatically, replacing existing method of manually generating documents
- Maintained school administrative system in Spring MVC Web framework and established functions for integrating academic performance to establish transcripts
-
July 2016 ~ Dec 2016K2J Studio
Back-End Developer Intern- Revitalized visual design of Miramar Cinema advertisement, implementing responsive web design(RWD) and incorporating multiple UI features to improve accessibility and functionality
- Maintained system of Cigna Taiwan Life Assurance Company Ltd and built API to scrape information from other websites automatically in ASP.NET MVC Framework
“The Cat is Out of the Bag” Explainable Risk Ranking with Financial Reports
The AAAI-20 Workshop on Knowledge Discovery from Unstructured Data in Financial Services. Workshop Papers New York, NY
Abstract
We propose an eXplainable Risk Ranking (XRR) model that uses multilevel encoders and attention mechanisms to analyze financial risks among companies. In specific, the proposed method utilizes the textual information in financial reports to rank the relative risks among companies and locate top high-risk companies; moreover, via attention mechanisms, XRR enables to highlight the critical words and sentences within financial reports that are most likely to influence financial risk and thus boasts better model interpretability. Experimental results evaluated on 10-K financial reports show that XRR significantly outperforms several baselines, yielding up to 7.4% improvement in terms of two ranking correlation metrics. Furthermore, in our experiments, the model interpretability is evaluated by using finance-specific sentiment lexicons at word level and an annotated reference list at the sentence level to examine the learned attention models.
Personalized TV Recommendation: Fusing User Behavior and Preferences
Submitted to Recsys, 2020 Preprint Online
Abstract
In this paper, we propose a two-stage ranking approach for recommending linear TV programs. The proposed approach first leveragesuser viewing patterns regarding time and TV channels to identify potential candidates for recommendation and then further leveragesuser preferences to rank these candidates given textual information about programs. To evaluate the method, we conduct empirical studieson a real-world TV dataset, the results of which demonstrate the superior performance of our model in terms of both recommendationaccuracy and time efficiency.
-
Developed full-stack system with Response Website Design layouts by using React.js
Design NoSQL database schema with Cloud Firestore and fetched desired restaurants’ information by integrated Yelp Business API, and Google Map API
To check out the introduction click here.
Our code is publicly available here
-
Built deep neural network architecture—XRR—for risk ranking with financial reports, yielding up to 7.4% improvement according to two ranking correlation metrics
Adopted multilevel attention mechanism to highlight crucial information at word and sentence levels, essential for finance applications
Determined using post-event return volatility as a risk proxy for such text analytics tasks, proving appropriateness of proxy for tasks in experiments
Evaluated learned attention models by utilizing finance-specific sentiment lexicons at word level and an annotated reference list at sentence level
To read the paper click here.
Our code is publicly available here
-
Conducted empirical studies on a real-world TV dataset and formally defined the problem of TV Rec
Proposed a two-stage ranking approach for TV Rec which consists of two components: one to model viewing behavior and the other for viewing preferences.
Demostrated our proposed approach consistently outperforms other baseline methods; more importantly, our two-stage ranking approach is parameter-free and efficient at inference, making it applicable and practical to real-world TV Rec scenarios.
To read the paper click here.
-
Developed full-stack social networking platform for users to initiate exercise groups in real-time by using Node.js, express, Firebase RealTime database, Facebook Login, and Google Map API
Applied Socket.io to create chat rooms for users in each exercise group to evaluate experience
Our code is publicly available here
-
Proposed wireless network protocol using 3-way authentication, 2nd-party inspection, and RSA algorithm, ensuring complete data packet transport and avoiding double spending
Created BLE wireless mobile payment system, Beepay by using Google BLE4.0 API
Developed a back-end system platform for displaying transaction records or deposits and Streaming APIs by using Servlet & JSP and MySQL
Contact Me
Swing by for a cup of coffee, or leave me a message:
