Wilson Wang Email & Phone Number

Taipei, Taipei City, Taiwan

• Led the development of an LLM-powered service, implementing a robust prompt engineering pipeline with Haystack and FastAPI. Integrated OpenTelemetry, Google Cloud Trace for comprehensive monitoring of usage, costs, and.
• Designed and implemented an LLM batch prediction framework using Airflow’s dynamic task mapping and deferrable operators, optimizing scalability for large-scale user segmentation with Trino. Successfully integrated.
• Led end-to-end development of a scalable error handling microservice supporting 7 connection types (S3 bucket, S3 file, GCS, SFTP, BigQuery, Google Sheets). Delivered Airbyte API integration and successfully deployed.
• Resolved a longstanding pod creation issue in Airbyte data integration that previously required manual restarts. Setup OpenTelemetry metrics collection to proactively alert on failed connection creation, enhancing.
• Designed and implemented integration across 6 feature components using the Feature Flag Decorator pattern, ensuring code maintainability and enabling ease of unit testing while accelerating feature delivery.
• Designed offline event segmentation and data preprocessing workflows leveraging Celery, Trino, and GraphQL. Accelerated feature delivery by 60% and achieved 100% key results completion, including custom segmentation.

• Decoupled the customer data extraction and loading workflow from Airflow by using Airbyte. Reduce the complexity of the DAGsand the CPU usage of Airflow by around 20%.
• Built the ETL pipelines that extract data from various sources, load them into the GCS and send them to BigQuery for further analysis.

• Improved the performance of an embedding lookup module based on double array trie and refactor it as a standalone library
• Implemented a webhook service receiving and processing data from Shopify webhook API, then sending them to the data store by using FastAPI and Redis
• Developed a python package template including environment setup, linting, formatting, testing commands, and GitLab CI/CD pipelines for better development experience and code quality

Taipei City, Taiwan

• KKTIX Cold Start Recommendation
• Developed a recommendation algorithm for KKTIX, the online ticketing platform under KKBOX Group
• DesignedPreference-Content Convolutiongraph learning algorithm and validated it through Query-Basedexperiments
• The precision in offline testing is increased by 2.5 times and it is deployed and launched in August 2019

Taipei City, Taiwan

Taipei City, Taiwan

* Improved support engineer working process through building automated scripts* Implemented multilingual sentences clustering to improve language translation quality

* Modularize the legacy code to be more maintainable* Refactor class to function components and Hooks in ReactJS

Taipei City, Taiwan

* Developed CMS user interface including form validation, image uploader, article lists, etc.* Developed RWD layout for client website* Collaborate with backend developers to design and implement RESTful API and Graph AP

Master'S Degree, Computer Science

Bachelor'S Degree, Computer Science

Bachelor'S Degree, Geometry/Geometric Analysis