━ About · 關於

Kasa

Backend Software Engineer

Project Experience · Selected Work

EZLO: Bitfinex Auto-Lending Platform (Personal Side Project)

Personal Project Live product ↗

A Bitfinex auto-lending platform I designed, built, and run myself. Pricing anchors to market depth; order placement is idempotent, with circuit breaking and full audit and monitoring. It started as a Python/FastAPI + Vue prototype, then I rewrote it end to end in Spring Boot + Next.js for stability and easier maintenance.

Key Responsibilities

  • Sole full-stack owner: backend, frontend, database, deployment, and production ops
  • Designed the market depth-anchored lending pricing strategy and order engine
  • Built the reliability and security side: key encryption, audit logging, notifications, monitoring
  • Set up containerized deployment and CI/CD so the service runs on its own

Technical Implementation

  • Backend in Spring Boot 3 / Java 21: RESTful API with Spring Security JWT and TOTP two-factor auth
  • Frontend in Next.js / React + Tailwind CSS + shadcn/ui, statically exported and deployed on Cloudflare Pages
  • Lending engine runs on a scheduled cycle with periodic reconciliation, using a write-DB-before-order idempotent design to avoid duplicate placements
  • Depth-anchored dynamic pricing against the funding order book, guarded by rate sanity checks
  • Integrated Bitfinex REST v2 (HMAC-SHA384) protected by a Resilience4j circuit breaker and a Redis token-bucket rate limiter
  • PostgreSQL (Flyway-versioned) + Redis (cache / distributed lock / rate limiting); user API keys stored AES-256 encrypted
  • Email (Resend) and Telegram notifications for earnings digests and system alerts
  • Tamper-evident audit log via a hash chain; monitoring and alerting with Prometheus + Grafana + Alertmanager
  • Multi-stage Docker builds + Cloudflare Tunnel reverse proxy; GitHub Actions CI/CD with secrets managed in 1Password

Results & Impact

  • The engine runs fully automated; idempotency and reconciliation mean no duplicate orders and a self-heal back to a consistent state when something fails
  • Circuit breaking and rate limiting isolate exchange-side outages, so the service stays up when the exchange doesn't
  • Structured logging with Prometheus/Grafana makes issues fast to pinpoint
  • Grew from a one-person script into a layered, testable multi-module platform

Backend Engineer · Payments & Gaming Platform Integration

2023/10 - Present

On the engineering team, integrating third-party payment gateways and gaming platforms and optimizing the backend, mainly to keep payments stable.

Key Responsibilities

  • Developed and integrated third-party payment gateway APIs in Java Spring Boot
  • Integrated gaming platforms and improved API efficiency and data sync
  • Optimized the backend admin system for speed and stability

Technical Implementation

  • Integrated several third-party payment APIs, reaching a 99.9% payment success rate
  • Cut average backend request processing time by 30%
  • Supported other teams, resolving issues quickly and improving product workflows

Results & Impact

  • More stable payments and a better checkout experience
  • Cut issue reports on payment and gaming interfaces by 50%
  • Made the backend easier to maintain and scale

C# Engineer · POS & BI Systems

2022/09 - 2023/10

C# engineer working on POS systems, the backend admin system, and BI reports, keeping things stable and improving them over time.

Key Responsibilities

  • Built POS systems in C# and integrated various peripheral devices
  • Built backend systems in .NET Core for scheduled tasks and data management
  • Built BI reports to visualize data for decisions
  • Maintained and improved existing systems, resolving user issues quickly

Technical Implementation

  • Built scheduled-task modules that improved automated data processing
  • Integrated BI tools and built 20+ report types for business decisions
  • Kept existing systems running stably for over a year with the team

Results & Impact

  • Made report queries 40% faster
  • Cut POS issue reports by 30%
  • Improved cross-team collaboration and noticeably sped up development