• Maintained and migrated existing ETL jobs to ensure near real-time and daily transfer of 10 million rows of data between Salesforce, Snowflake, and Amazon S3, using Python, AWS Step Functions, Lambda, Glue, and EventBridge.• Designed and implemented a new microservice ETL system, increasing capability by 10x to onboard and process multiple data sources, using SOQL, SQL, Pandas DataFrame, and REST API.• Provided production support to resolve data anomalies for 100+ Business Analysts, using Python, Pandas, PySpark DataFrame, CloudWatch, SOQL, SQL, and Splunk.• Implemented automated acceptance and functional tests for new features, using Python Behave, Jenkins Pipeline, and Jira X-ray, enabling continuous integration.• Improved unit test coverage to over 95%, ensuring product compliance with enterprise standards and providing flexibility in test coverage requirements for new features, using Python.• Developed Infrastructure as Code (IaC) to manage instances and resources for the microservice system, using AWS CDK and CloudFormation.• Enhanced error handling and logging to reduce debugging and trace-back time, using Python and AWS CloudWatch.• Automated AWS deployment in the development environment, achieving a 10x improvement in deployment speed, using Linux shell scripts.• Documented step-by-step procedures in Confluence to streamline knowledge transfer and reduce onboarding costs.

• Enhanced efficiency and customer satisfaction for 25+ franchises and 5000+ locations with an automated voice-based service for 20,000+ restaurant orders daily. Utilized Scala, Clojure, Quill (for SQL), and Circe (for JSON). Deployed on GCP, integrated voice recognition, synthesis, and NLP services from Microsoft Azure.• Reduced voice recognition costs by 40%, potentially saving $144,000 annually, by developing a new RESTful audio stream client for Speech Recognition Service on Microsoft Azure, using Scala.• Developed a RESTful ToIP Text Service channel, resulting in a 40% increase in new business opportunities, using Scala, Twitter Finagle, SQL, and Circe.• Achieved a 100% recovery rate of unanswered RESTful API calls by developing a Retryable Filter for RESTful calls within the SMS client, using Scala and Twitter Finagle.• Prevented 75% of user call drops by implementing an order resuming capability feature, using Scala, SQL, and JSON.• Reduced response time by 50% with load balancers for voice synthesis, using Scala and Twitter Finagle.• Developed a hash string generator to efficiently map orders and services, enabling shortened URL payment applications, using Scala and SQL.• Lowered NLP failures by 30%, improved various voice command handling, and enhanced exception handling using Scala, leveraging insights from error logs in Datadog and Grafana.• Improved data retrieval accuracy by 30% for address parsing, using Scala and Regex(Regular Expressions).• Improved UI sections and implemented keyboard shortcuts for testing webpage UI, using Clojure and HTML. Achieved a 30% increase in testing efficiency.• Managed Kubernetes cronjobs, and optimized loading efficiency by 10% through YAML and JSON.

• Reduced assessment time by 75% and eliminated human assessment through an Online Judging system deployed on Microsoft Azure, using Python, Django, PostgreSQL, Redis, and Linux Shell.• Saved 56 hours per week and reduced operational costs by 33% by automating Docker container processes using Linux Shell scripts.

Free5gc department of NCTU• Developed a verification system to validate JSON messages in 5G core network, reducing validation time by 70%, using TTCN-3 and C++.• Identified and resolved 15 critical issues in the 5G core network, using Wireshark, C++, Postman, and CURL.• Enhanced AMF network function responsiveness by refactoring 1000+ lines of code, using a combination of C and C++.• Reduced testing time by 50% with simplified Git usage and process launching using Linux Shell scripts and MAKE files.

Ministry of Transportation and Communications TW & NCTU • Developed an embedded system for data reading from taxi meters, reducing data loss by 90%, using Arduino and C++.• Developed a data retrieval system, improving data completeness by 95%, using VBA.• My product was produced 50+ times, and collected and processed 10,000 GPS data points from 1000+ taxis.