Software Developer
Current- Designed a low-latency service achieving a P99 response time under 100 ms, without relying on caching, for crawler systems, or other in-house tools.- Optimized system performance, achieving an impressive 82.5% reduction in P99 response time. Simultaneously, increased qps from 70 (utilizing 4 workers) to 200 (with 1 worker), marking a substantial 185% improvement in QPS without cacahing.- Built an API gateway for products, which involved implementing caching, authentication, authorization, and rate limiting strategies.- Created the end-to-end design of a product, serving as the cornerstone for numerous successful products, and iteratively improved and optimized it to meet evolving user needs.- Collaborated with algorithm engineers to establish multiple ETL pipelines, enhancing data processing capabilities by parallelizing data pipelines within limited resources.- Established a GitLab CI/CD architecture for algorithm engineers, optimizing development workflows, achieving a 73% reduction in image build time, and ensuring seamless code integration in Kubeflow and Cronjob.- Developed an extensible monitoring system, allowing easy addition of various metrics and data for diverse purposes in Prometheus. Reduced system diagnosis time by 80% and consolidated statistics through data visualization in Grafana.- Conducted and enhanced AI algorithms by implementing a Celery worker, resulting in a 73% reduction through adjustments in worker settings.- Managed an AWS RDS cluster, monitoring multiple metrics to fine-tune performance and optimize cost-effectiveness, resulting in a 10% savings.python, nodejs, golang, kubernetes, awsframework: fastapi/celery/nestjs/kubeflow
