Derek(Tsai-Yun) Hsieh

SWIII - CIS Research Assistant, WPC Department of Information Systems, Faculty Support XLVI • Designed and validated analytics workflows on large experimental datasets using AWS SageMaker, achieving 94% accuracy. • Conducted deep-dive and ad-hoc analyses to evaluate data patterns, variability, and reliability, supporting business-aligned insights. • Conducted statistical analysis and hypothesis testing to assess model performance and data integrity. • Built automated data ingestion and preprocessing pipelines (Python, Selenium, MySQL) to ensure data availability and consistent analytical inputs. • Resolved data quality and process issues through root-cause analysis, ensuring reliable and reproducible insights. • Produced technical documentation and validation artifacts to support auditability and system maintenance.

Global Disaster Risk Detection and Analytics System (ASU <> Avnet Capstone Collaboration Project) •⁠ ⁠Collaborated on building an AI-driven platform integrating real-time alerts, news analytics, and geospatial risk scoring to strengthen supply chain resilience and crisis response planning. •⁠ Created data-driven metrics and visualization frameworks that incorporated risk evaluation at a regional level for strategic logistics decision-making.

The primary goal at MUFG was to improve user experience by enabling quicker access to the right data domain through enhanced keyword search functionalities. By harnessing the power of the RoBERTa NLP model, our team made significant strides, increasing domain classification accuracy from 68% to an impressive 90%. Additionally, we crafted a user guide to streamline search processes, ensured consistent terminology usage across departments, and seamlessly integrated with Collibra to enhance AI tool utilization in searches. • Improved Collibra search engine with RoBERTa NLP model, boosting performance 20%. • Checked Collibra workflows to ensure controlled updates, rule compliance, and correct system behavior. • Enhanced Stewardship keyword search via GPT optimization, increasing accuracy 25%. • Refined RoBERTa search models and documented compliance, improving search relevance by 20%. • Analyzed and standardized unclear data domain definitions to improve data quality and reduce misinterpretation.

When I was working as a data analysis intern as an engineer in Hsinchu Lioneers professional basketball team, I helped the company integrate multiple data (ticketing systems, merchandise systems, investments) to find target customers and allow the marketing department to use the integrated data to plan and meet customer needs. Through the lowest cost, we create the most excellent revenue. To this day, the department is still optimizing and updating data to launch better activities through the system I built at that time, and the revenue has grown significantly year after year. • Processed client data using Excel VLOOKUP and MySQL, improving marketing campaign effectiveness by 170%. • Developed customer journey mapping and tagging strategies using SAP ERP modules, increasing high-value customer spending by 120% per quarter. • Conducted customer segmentation using Tableau RFP model, leveraging data visualization and predictive analytics to improve sponsor engagement by 40% and raise project revenue by 28%. • Partnered with CRM/CDP vendors and cross-functional teams to integrate APIs, enhancing data accuracy and driving a 56% increase in online consumer engagement. • Streamlined data systems, supplier reports, and customer feedback to produce QLTC reports and maintain analytics systems, improving supply chain management.

The data analytics intern is mainly responsible for the integration and analysis of social media data and through optimization to help the company provide better services and communication capabilities for the social media of the T1 League Basketball Professional League. This not only increases the number of social media tracking and interactions but also results in online interactions being actually converted into offline people, who are motivated to purchase tickets to watch games allowing the company and fans to achieve a win-win situation. • Analyzed customer behavior on social media and converted online engagement into offline actions, increasing CVR by 12%. • Utilized text mining to investigate market trends, delivering cross-departmental project recommendations that boosted customer engagement by 25%. • Applied moving average analysis to social media data, optimizing project lifecycle and increasing post reach by 21%.

Introduction of iFITTIN: An APP, using food identification ways via deep learning to help people record and manage their diets. Supported R&D deployment and pre-release execution planning by coordinating deployment readiness activities, tracking execution progress, and ensuring alignment with defined project timelines and deliverables. • Applied DMAIC methodology to define data quality requirements and deployment validation criteria, assessing system stability, variability, and readiness prior to production release. • Designed and executed pre-deployment validation and risk control processes, identifying potential data and workflow failure points and supporting corrective actions to reduce post-deployment operational risk. • Enabled automation-driven deployment workflows using AWS S3 and Rekognition, reducing manual processing effort by 60% and supporting stable adoption across 5,000+ internal users.

As the brand general coordinator of The Dealer, I am mainly responsible for coordinating human and financial resources, completing a huge project at the lowest cost, analyzing customer groups and communities, and writing proposals to attract investors willing to sponsor events. In the end, we managed to increase the scale of the event and rank among the top three student music festivals in Taipei. We increased the number of Instagram followers by 413% and FB followers by 120%. During the event, we held two music festivals and three music lectures. It has grown ten times compared to other years and has raised more than US $6500 in activity funds. Serving as the CEO of this organization made me realize that using data analysis technology can help companies successfully hit their target customers with the least cost and make companies successfully profit. • Utilized data-driven insight and problem-solving skills to establish a music festival business. • Applied Python (Beautiful Soup) and CKIPTAGGER for text mining and keyword analysis, identifying high-value band lineups that resulted in a 35% increase in ROI. • Leveraged A/B testing to increase Instagram followers by 413% and Facebook followers by 120%, boosting CTR by 8% in four months. • Converted online traffic into 3,000+ attendees, establishing the event as one of the top 3 student music festivals in Taipei.

Position: Model Entrepreneur- Model Entrepreneur- Software Engineer for VIBSER • Acquired two merit awards in prestigious national entrepreneurship competitions. • Developed a college-oriented, interest-based social networking App. • Utilized Pearson correlation and linear programming for insights from 200+ survey responses. • Developed a basic matchmaking feature using Python's Flask Framework. • Connected our matchmaking feature with LINE Bot Channel via Messaging API SDK and Webhook. • Successfully delivered a Minimum Viable Product for our business model.