About Me
I help product and engineering teams turn ambiguous requirements, cross-role communication gaps, and production constraints into practical backend / platform direction. My strength is identifying the real needs behind written or verbal requests, clarifying responsibilities across roles, and translating technical trade-offs into language that Product, UED, PM, Frontend, Backend, Mobile, and Platform stakeholders can act on.
As a Senior Backend / Platform Engineer, I can own backend services from system design to production operation: defining service boundaries, designing authentication / authorization and API flows, standardizing reusable integration patterns, improving latency and reliability, and strengthening CI/CD and observability. I work best in environments that need both technical direction and team alignment, where backend architecture, platform capability, cross-team execution, and production reliability must move together.
Experiences
- Owned backend / platform services for uniopen, PCSC’s high-traffic web portal and mobile app serving 3.5M+ users and ~10K peak RPS on AWS + EKS, ensuring reliable authenticated user journeys during campaign and push-notification traffic spikes of ~200K users.
- Designed and built the company’s core post-login authorization service from 0 to 1, defining service boundaries, API access validation, token handling, and identity lifecycle management for business-critical membership and authorization flows.
- Designed reusable OIDC/SSO integration patterns through authentication SDKs and API service templates adopted by ~20 backend services, reducing integration / bootstrap effort by ~80% while improving consistency across teams.
- Led performance baselining, load testing, and tuning for authenticated authorization services, reducing critical endpoint latency by up to 77% and increasing throughput from 300 to 500 RPS.
- Investigated and resolved a complex multi-service production incident during traffic spikes, reducing DB CPU from >85% to 20–30%, eliminating timeouts, and lowering operational risk for login-dependent flows.
- Standardized IaC-based CI/CD workflows, CloudWatch dashboards/alarms, centralized logging, and audit logging across Spring Boot and Python backend services, cutting pipeline time from 40 minutes to 20–25 minutes and reducing incident investigation time by ~30%.
- Designed and operated a shared redirect service across web and mobile entry points, supporting campaign routing, brand traffic, SSO handoff, Kafka-based asynchronous event delivery, and click tracking for hundreds of pages, dozens of brands, and millions of product cards.
- Tech:
- Languages: Python, Java
- Frameworks: FastAPI, Spring Boot
- Datastores: PostgreSQL, Redis, DynamoDB, DocumentDB
- Cloud/Infra: AWS (EKS, S3, SQS, SNS, Lambda, CloudWatch, MediaConvert)
- Tooling: Terraform, Helm, Docker, Kubernetes, Istio
- Testing/Perf: k6, Vegeta
- Owned backend APIs for a 100K+ user online learning platform with ~4K DAU, supporting revenue-critical and retention-critical workflows across cart, payment, learning progress, membership, revenue sharing, coupon systems, and internal operations.
- Modernized a Laravel monolith from v7 to v10 in 2 months through incremental migration, compatibility validation, and unit-test coverage, reducing legacy framework maintenance and security risk while improving API p99 latency by ~20%.
- Improved database performance for high-frequency purchase, learning, and back-office workflows through request-path profiling, EXPLAIN analysis, index tuning, and schema refactoring, reducing key query time from 10–20 seconds to ~1 second and improving responsiveness of core user and operational flows.
- Operated the AWS production stack across ALB, EC2, and RDS, covering CI/CD, autoscaling, CloudWatch dashboards/alarms, load testing, release verification, and post-release monitoring to improve release safety and service stability under traffic peaks.
- Improved production monitoring and incident response with Slack alerting, reducing abnormality detection time by ~80% and shortening issue detection time for payment, learning, and internal operational flows.
- Built asynchronous backend workflows with SQS + Lambda for LINE notifications and personalized recommendations, supporting user revisit, course engagement, and personalized outreach while contributing to a ~20% retention improvement.
- Tech:
- Languages: PHP, JavaScript, Python
- Frameworks: Laravel, Lumen, Vue, Nuxt
- Datastores: MySQL, SQLite
- Cloud/Infra: AWS (EC2, ALB, RDS, SQS, Lambda, CloudWatch, ECR, S3)
- Tooling: Docker, Docker Swarm, OpenAPI/Swagger
- Testing: PHPUnit
Skills
Backend Engineering
Cloud & Infrastructure
Databases & Caching
Platform, Reliability & Observability
Architecture & Integration
Tools
Projects
Side projects and tools.
(RUST) Documentation Embedding
- Rust CLI for document embedding and Qdrant-based semantic search, demonstrating vector search, CLI tooling, and AI infrastructure experimentation.
(Python3) ChatGPT Token Usage Calculator
- Python utility for estimating prompt token usage before API calls, supporting cost-aware LLM workflow design.
(Docker) Dockerize Templates
- Reusable Docker templates for common backend stacks, standardizing local development setup and reducing project bootstrap effort.
Certifications
Validates core AWS architecture skills across networking, compute, storage, security, and reliability best practices.
OSS Contributions
Open source contributions.
actix/examples
- Community examples showing Actix ecosystem usage.
openai-cookbook
- Examples and guides for using the OpenAI API.
design-patterns-for-humans-zh-TW
- Traditional Chinese translation of design patterns.