Full Stack Developer
Full Stack Developer with expertise in end-to-end system design and implementation. Adept at quickly adopting new technology frameworks and solving complex business problems with independent project delivery capabilities. Particularly experienced in integrating AI technologies with business applications, and committed to balancing code quality with user experience.
Expert in Go language and its ecosystem, proficient in Java development, mastery of MVC architecture patterns and RESTful API design
Expert in React ecosystem (Next.js), proficient in TypeScript, HTML/CSS, with experience in WeChat Mini Program development
Expert in ComfyUI workflow design and development, mastery of AI model deployment and optimization, proficient in RAG technology and vector knowledge base construction
Proficient in PostgreSQL and MySQL for relational database design and optimization, mastery of MongoDB document databases and Redis caching strategies
Proficient in Docker containerization, CI/CD pipelines, Linux system operations, and network programming
Familiar with microservice architecture, message queues, load balancing, and distributed system design principles
Full Stack Engineer|Guangzhou, China
As a core technical team member, I independently manage the entire development lifecycle of AI color change applications from system design to deployment:
Java Developer|Platform Division|Guangzhou, China
Served as a core backend developer in the data middle platform team:
Built an intelligent vehicle color change visualization platform to help auto refinishing shops increase sales conversion rates while meeting consumer experience demands.
Traditionally, customers couldn't visualize vehicle customization outcomes before implementation, with information barriers between suppliers and shops
Designed Stable Diffusion-based adaptive vehicle color change workflows combined with ControlNet for precise control
Achieved high-quality color preview generation within seconds, supporting 60+ colors and 95%+ of common vehicle models
AI processing tasks are computation-intensive, causing head-of-line blocking on single machines
Designed Kafka-based message queue with dynamic scheduling for intelligent task distribution and load balancing
3x improvement in system throughput, 60% reduction in average wait time, 99.5% success rate for automatic retry of abnormal tasks
Traditional FAQs couldn't meet personalized consultation needs
Built vector knowledge base using RAG technology integrated with large language models for intelligent Q&A
Significantly improved customer self-service capabilities, reducing human customer service workload
B2B clients needed independent management of permissions, resources, and data
Designed horizontal partition multi-tenant architecture with data isolation and resource quota management
Supported horizontal service scaling, improved resource utilization by 45%, and significantly enhanced system scalability
Needed to acquire users cost-effectively while improving retention
Designed membership tiers, referral rewards, points system, and social sharing features
Reduced user acquisition costs by 35%, improved average retention by 28%, 300% increase in monthly active users
Provided data developers with comprehensive capabilities for data quality assessment, monitoring, and governance to ensure accuracy, completeness, and consistency during data development.
Performance bottlenecks in large-scale data comparisons affecting development efficiency
Rewrote core algorithms in Rust, optimized data structures, reducing time complexity from O(nlogn) to O(n)
52% improvement in API response speed, significant reduction in CPU and memory usage, supporting rapid comparison of tens of millions of records
Need for flexible configuration of diverse data quality validation rules
Designed expression-based rule engine supporting complex condition combinations and custom functions
Implemented 30+ data validation rules covering over 90% of common data quality issues
Ensuring sensitive data security during development and testing
Designed prefix tree-like structures for efficient masking with multi-level strategies and permission controls
Met compliance requirements while maintaining data usability, with 65% improvement in processing efficiency
Built unified performance evaluation standards by integrating data from multiple business systems to support HR decision-making.
Needed to integrate differently structured data from CRM, Zentao, DingTalk, and other systems
Designed and implemented an Extraction Data Layer (EDL) with metadata management for unified mapping
Successfully integrated 5 key system data sources with automated updates, reducing manual data organization work by 90%
Difficulty in uniformly quantifying and comparing performance metrics across different positions
Adopted star schema with fact and dimension tables to build performance data models
Established 15 core dimensions and 8 key fact tables supporting multi-dimensional performance analysis, reducing user analysis time by 65%
Varying source data quality affecting performance evaluation accuracy
Implemented a Data Warehouse Detail (DWD) layer with data quality monitoring metrics and exception handling processes
Improved data accuracy to 99.5% with automatic anomaly detection rate exceeding 90%
Need for flexible definition and calculation of complex performance metrics
Developed an expression-based metrics calculation engine with Data Mart (DM) layer customization
Supported automatic calculation of 200+ performance metrics, meeting the customized needs of different departments
Independent Developer