VC Vinson.Cheng WUNZACK
Open for Work // Taichung / Remote

Senior Backend Engineer / Tech Consultant

Vinson.Cheng WUNZACK

High-concurrency backends that stay stable, fast and maintainable

Go · C# | Kafka · K8s · GCP | Event-Driven × Data Consistency

Talk about your project View case studies

Services

How I Can Help

From architecture design to production tuning — owning the technical decisions on critical backend modules.

01

High-Concurrency Architecture

Design horizontally scalable microservices handling tens of thousands of CCU and tens of millions of daily events, balancing performance and maintainability.

02

Cloud-Native Deployment

Containerized deployment on GCP / GKE / Cloud Run, with CI/CD, observability and cost-performance tuning.

03

Event-Driven & Data Consistency

Kafka, Outbox/Inbox, idempotency and distributed locks to keep cross-service transactions exactly-once under heavy load.

04

Integration & Refactoring

Unify outbound APIs, decouple systems, and refactor legacy code for maintainability and stability.

02 / Selected Work

Proven Results

Complex requirements. Working systems.SELECTED ENGINEERING / 01—04

CASE STUDY / 01

Real-time · Distributed platform

From zero to a real-time platform in 91 days

01 / Problem

An early-stage interactive platform needed production architecture, microservices, delivery pipelines, observability and a team ready to operate it.

02 / Approach

As technical co-founder, designed 13 microservices for real-time interaction, state synchronization and resource coordination. Connected services with WebSocket and gRPC; used event-driven processing, Outbox and idempotency for data consistency. Built an OBS WebSocket Bridge control console combining venue management, scheduling, broadcast and audio controls, quality monitoring, session operations and incident intervention. Introduced K8s GitOps, observability, reusable service scaffolds and AI-assisted delivery, while enabling independent team operations.

03 / Impact

91 daysconcept to launch
13microservice repos
1,591Git commits
420K+net lines of code

Technology

  • Go
  • gRPC
  • Kafka
  • WebSocket
  • OBS WebSocket
  • K8s
  • ArgoCD
  • OpenTelemetry
CASE STUDY / 02

Digital services · B2B platform

Millions of events. One resilient platform.

01 / Problem

A multi-integrator campaign platform needed to support tens of thousands of concurrent users and millions of daily events while keeping shared resources, campaign budgets and cross-service data consistent.

02 / Approach

Built event-driven microservices with Kafka, using Redis distributed locks and atomic Lua operations to coordinate shared state and campaign budgets. Unified integration APIs and refactored synchronization scheduling and token-expiry cleanup.

03 / Impact

×40Shared-state sync performance
×10Token-expiry cleanup perf
7–12MKafka events per day
10,000+Peak concurrent users

Technology

  • Go
  • Kafka
  • Redis / Lua
  • MongoDB
  • GCP
  • K8s
CASE STUDY / 03

Manufacturing · Core Systems

Manufacturing Core Systems Integration & Refactor

01 / Problem

SCADA / MES / ERP(Baan) and Fanuc equipment were fragmented and tightly coupled; failed transactions were hard to recover and the legacy code was hard to maintain.

02 / Approach

Integrated SCADA / MES / ERP with Fanuc equipment; introduced transaction idempotency and error-recovery; refactored the legacy architecture for maintainability and stability.

03 / Impact

IdempotentTransactions & error recovery
4 systemsSCADA/MES/ERP/Fanuc integrated

Technology

  • C# / .NET
  • SCADA
  • MES
  • ERP
  • Fanuc
SCADA, MES, ERP and Fanuc connected to an integration layer supporting idempotency and error recovery.
Architecture illustration · not a client screenshot
CASE STUDY / 04

AI Agent · Household shopping × Evaluation

Household Agent: from conversation to everyday action

01 / Problem

Shopping needs, lists and replenishment habits are scattered across everyday conversations. The product needs to follow multi-turn context, clarify missing details and turn intent into confirmable list actions and reminders, while tracking quality changes across models and prompts.

02 / Approach

Built household-agent with a mobile-first Vue 3 PWA, Go services and an admin console connecting daily lists, replenishment reminders and a purchasing agent. Structured conversation choices clarify intent; tool calls pass schema validation, workspace authorization and application services with traceable records. PostgreSQL, Kafka and a transactional outbox support event consistency. The admin console compares SKILLS, TOOLS, CHITCHAT and MULTITURN evaluations, including scores, regressions, costs and failures, to assess model and prompt changes.

03 / Impact

4 suitesSkills, tools, chat and multi-turn evaluation
183 casesEvaluation set shown in the screenshot
91.59One displayed run average, not overall accuracy
TraceableModel differences, regressions and run status

Technology

  • Go
  • Vue 3 / PWA
  • PostgreSQL
  • Kafka
  • Redis
  • Tool Calling
  • Agent Evaluation
  • Kubernetes

Tech Stack

Tools & Technologies

$ cat lang // Languages
  • Go
  • C# / .NET
$ cat arch // Architecture
  • Microservices
  • Event-Driven
  • DDD
  • Outbox / Inbox
  • Idempotency
$ cat data // Messaging & Data
  • Kafka
  • MongoDB
  • Redis
$ cat cloud // Cloud & Deploy
  • GCP / GKE
  • Cloud Run
  • Kubernetes
  • Docker
  • CI/CD
  • ELK

About

About Vinson.Cheng

8+ years of backend development across manufacturing and digital interactive services, from factory equipment integration to platforms serving tens of thousands of concurrent users.

I independently own architecture design and technical decisions on critical modules, specializing in cloud-native deployment, high-concurrency systems and distributed data consistency. I like turning messy business logic into clear, testable, long-lived systems.

Contact

Let's talk about your project

Freelance, consulting or long-term — drop your needs and I usually reply within 24 hours.

Or reach me directly jinn713212@gmail.com