Chan Shun-Chih (詹舜智)

Chan Shun-Chih

詹舜智 · shunzz

AI-Native Software Engineer · Senior Systems Analyst · Taipei, Taiwan

Four years of Python backend and AI application engineering. I specialize in spec-driven development with AI agents (Claude Code, GitHub Copilot CLI; the open-source superpowers workflow, tuned per project), independently delivering enterprise systems, ML models, cloud IaC and consumer products — then institutionalizing the workflow through internal talks, reusable Claude Skills with evals, and a self-built MCP Gateway serving as company-wide AI infrastructure. Previously led two AI video-generation products from zero to 7,000+ users, halving monthly infrastructure cost ($40K → $20K) and cutting failure rate from 30% to under 1%. JLPT N1.

Three layers

  1. Ship products with AI

    Enterprise projects and side projects, with speed and quality in the git history: an allocation platform in 2.5 months, an MCP Gateway live in 3 days, a 62-test app in one day.

  2. Manage how AI works

    Spec first, TDD, an automated AI code-review loop, prompt governance; one shared AGENTS.md driving both Claude Code and Copilot CLI.

  3. Build tools for AI

    A self-built MCP Gateway, Claude Agent Skills with evals, project-specific skills (e.g. an Azure DevOps PR flow) — and internal advocacy.

Experience

IT subsidiary of a retail group — Senior Systems Analyst (AI Application Engineer)

May 2026 – present · Taipei

Cloud architecture and AI application development across cloud architecture, DevOps, data engineering and AI products.

Retail allocation platform (primary project)

  • Independently delivered an enterprise retail allocation platform (1,100+ files) in 2.5 months with AI collaboration; heavy requirements interviews, allocation logic designed from per-store sales data and special cases
  • Backend Python 3.12 / FastAPI / SQLAlchemy 2 / Alembic / PostgreSQL in a layered architecture (api / core / engine / etl / repositories / services); frontend React 19 + TypeScript / Vite / TanStack Query·Table·Virtual / react-hook-form + zod / Tailwind 4
  • Contract-first: openapi.json → openapi-typescript generates frontend types, keeping both sides in sync
  • CI/CD on Azure Pipelines: PRs run backend pytest and frontend lint/test/build in parallel; 8 backend test modules, 56 frontend test files, UAT smoke tests
  • Requirement alignment via single-file HTML prototypes that stakeholders preview before real development, cutting the spec → build → acceptance cycle; disciplined docs (decisions / lessons / runbooks / roadmap)

Demand-allocation ML models

  • Store-level allocation forecasting with LightGBM, CatBoost, Optuna tuning and isotonic calibration
  • Systematic pipeline: baseline → feature engineering → ablation → blending → rules overlay → misallocation diagnosis; delivered ML allocation plan draft, ER model and table schema

Enterprise Customer Data Platform (6 repos / 2,200+ files)

  • React 19 + TS + TanStack Router/Query frontend, .NET 8 Web API, .NET 8 batch system with unit tests; SQL Server multi-schema design + Azure Synapse warehouse (CTAS / HASH-distributed tuning)
  • AI API: Python 3.13 + FastAPI + Google ADK agent pipeline + LiteLLM → Azure OpenAI; automated product tagging (migrated from a legacy service) and a 3-stage marketing-copy pipeline; strict API contract (four status codes, per-user rate limiting, fail-honestly-no-fallback)
  • Three environments (dev/test/prod) per repo on Azure DevOps pipelines; incident records and architecture runbooks

MCP Gateway (self-built AI infrastructure)

  • Wraps internal HTTP APIs as MCP tools so Claude Code / Codex / Copilot reach internal systems through one secure entry point; tool catalog in DB, approved tools go live without code changes or restarts
  • Security: per-client Bearer auth, SSRF protection (blocks localhost/cloud metadata + allowlist), Fernet-encrypted upstream credentials, audit logs with sensitive-parameter masking
  • Python 3.12 / MCP SDK (Streamable HTTP) / Starlette / SQLAlchemy 2 / Streamlit admin; 20+ test files incl. acceptance tests (pytest-asyncio + respx); v1 live in 3 days

Log anomaly analysis

  • Gradio UI with two-stage filtering of large logs → Azure OpenAI analysis → SQLite; strict three-layer decoupling (filter / ai / db); token and cost estimation before each AI call with threshold warnings

Enterprise chatbot product deployment

  • Source governance and dockerization of an enterprise chatbot product (Java/JSP/Kotlin, 700+ files): Compose with MSSQL 2022, Redis Stack, Solr 9.5, custom parser; DB init, health checks, fresh-VM deployment SOP; git filter-repo history cleanup, Git LFS

Azure IaC & data engineering

  • Modular Bicep IaC (network / keyvault / storage / postgres / appservice): IP allowlist + Private Endpoint dual path, Key Vault secrets, VNet Integration + RBAC, rollback docs
  • Dify (open-source LLM platform) on Azure Container Apps (PostgreSQL + pgvector / Redis / Storage, with cost estimate); Terraform for LLM resource groups; ADF pipeline analysis and design (auto-ordering / inventory / sales forecasting), ML pipeline retry/webhook design

AI engineering methodology (institutionalized)

  • One shared AGENTS.md driving Claude Code and Copilot CLI on enterprise projects; git worktree for parallel branches
  • Adopted and operated the agent-skill toolchain (open-source superpowers workflow) and built an automated AI code-review loop (AI fixes → commit → replies to review thread → re-review until it converges)
  • Packaged team domain knowledge as Claude Skills with evals (Azure Synapse tuning, ADF standards) plus project-specific skills (e.g. an Azure DevOps PR flow)
  • Internal tech talk on the requirement → AI prototype → user alignment → build workflow (14-page outline, script, 4 diagrams) with 5 reusable de-projectized skills
  • Python
  • FastAPI
  • React 19
  • TypeScript
  • .NET 8
  • PostgreSQL
  • SQL Server
  • Azure Synapse
  • Azure Data Factory
  • Azure Container Apps
  • Bicep
  • Terraform
  • Google ADK
  • Azure OpenAI
  • MCP
  • LightGBM
  • CatBoost
  • Azure Pipelines
  • Claude Code
  • GitHub Copilot CLI

OSENSE Technology — Software Engineer (AI Application Engineer)

May 2024 – Dec 2025 · Taipei
  • Led OVideo, an AI-driven content-to-video platform serving 1,800+ users, incl. membership, subscriptions, payment integration and Azure deployment
  • Designed the full video-generation pipeline: LLMs (GPT-4, Claude, Gemini) for content analysis, scripting and image-text matching; FFmpeg for processing, subtitles and encoding; async tasks with Server-Sent Events progress
  • Performance and cost: parallelized the pipeline to cut average generation time from 30 to 15 minutes (-50%); re-architected Azure services to halve monthly infrastructure cost ($40,000 → $20,000); serverless architecture scaled capacity from 10 to 100+ concurrent users; 1,100+ videos processed
  • Introduced Clean Architecture, cutting new-feature development time by 50%; designed a PostgreSQL / MongoDB hybrid for high concurrency
  • Built Osense AI, a LINE-based AI avatar and voice-cloning system (5,400+ users): lip-sync and voice cloning, full LINE Bot user flows; fixed upstream open-source bugs and error handling to cut failure rate from 30% to under 1%
  • Rapid PoCs with Gradio and light frontends for client demos; introduced AI tooling to speed up development
  • FastAPI
  • Celery
  • PostgreSQL
  • MongoDB
  • Azure (VM, Container Apps)
  • Docker
  • FFmpeg
  • Wav2Lip
  • TTS (Azure Speech, ElevenLabs)

Jung Shing International — Software Engineer (Backend)

Jul 2022 – Apr 2024 · Taipei
  • Built internal web systems with Django: table design from requirements interviews, API and flow planning; Django + jQuery integration with GitLab CI/CD
  • Developed and maintained scrapers with Selenium and Requests; optimized existing scrapers and anti-bot handling
  • Built a PDF invoice extraction tool that captures key fields into the database, reducing manual entry to a field check
  • Introduced GitLab version control and a basic development process; applied the Repository Pattern to isolate data access
  • Python
  • Django
  • Selenium
  • Requests
  • GitLab CI
  • jQuery
  • Linux

Side projects & contributions

Career Agent — AI job-search assistant

Jun 2026 – ongoing

Upload résumé → AI diagnosis → crawl listings → semantic fit ranking (score + reasons + missing skills) → tailored cover letters → application tracking; cloud multi-user and local personal-assistant architectures. FastAPI + MongoDB + Claude API + Playwright; React + TS + Mantine; 73 commits in 3 days, 10K+ views on the launch post, real users. Built end-to-end with the Claude Code + superpowers agentic workflow: I owned requirements, architecture, task breakdown and review.

Live GitHub

Suancui (酸脆) — Threads AI roast generator

2026

Viral-loop design (share to unlock hidden styles); TypeScript npm-workspaces monorepo: React frontend, Cloudflare Workers + Hono + KV, swappable scraping layer (self-hosted GraphQL crawler / third-party API fallback); four packages each with vitest tests; anti-fabrication and anti-harassment guardrails. 67 commits in one day.

Daily work-log system

2026

Flask + SQLAlchemy; full spec-driven flow (design spec → implementation plan → TDD), ~2,100 lines, 62 tests, built in one day.

iThome Ironman 2023: Django source-code deep dive

Sep – Oct 2023

A 30-day series tracing Django internals from everyday commands and APIs into request handling, the ORM and middleware.

Blog

Other

Reverse-engineered an undocumented Central Weather Administration API (typhoon storm-radius probability); Text-to-SQL experiments (Vanna + Wren Engine semantic layer); personal RAG agent (chromadb); 56+ posts on a Hexo tech blog.

Tech stack

Languages
  • Python (primary)
  • TypeScript
  • C# / .NET 8
  • SQL (SQL Server / Synapse / PostgreSQL)
  • Bicep / Terraform
Backend
  • FastAPI
  • Flask
  • Django
  • Starlette
  • Hono
  • SQLAlchemy 2 + Alembic
  • Celery
  • async (asyncio / asyncpg / motor)
  • RESTful API
  • SSE / WebSocket
Frontend
  • React 18/19 + TS
  • Vite
  • TanStack (Query / Table / Router / Virtual)
  • Tailwind
  • zod
  • Mantine
AI / LLM
  • Claude API
  • Azure OpenAI
  • GPT-4 / Gemini
  • Google ADK
  • LiteLLM
  • MCP (self-built server)
  • RAG / chromadb
  • Prompt engineering
  • Claude Skills + evals
  • FFmpeg / Wav2Lip / TTS
  • Token cost control
Cloud / DevOps
  • Azure (DevOps Pipelines, Bicep, Terraform, Synapse, Data Factory, Container Apps, Key Vault, Private Endpoint)
  • Cloudflare (Workers / Pages / KV)
  • Docker
  • Git LFS / git worktree
  • GitLab CI
  • uv / ruff
ML
  • LightGBM
  • CatBoost
  • Optuna
  • scikit-learn
Scraping
  • Playwright
  • Selenium
  • Requests / curl_cffi
  • BeautifulSoup
  • Anti-bot handling
Testing / architecture
  • pytest (asyncio / respx)
  • vitest + Testing Library
  • xUnit
  • TDD
  • UAT smoke / acceptance tests
  • Clean Architecture
  • Repository Pattern
  • Design patterns

Education

Workforce Development Agency — AI Cloud Web Application Development Program

Dec 2021 – May 2022

Full-time intensive, 6 months

Wenzao Ursuline University of Languages

Sep 2016 – Jun 2020

B.A. in Japanese

Languages

Chinese
Native
Japanese
JLPT N1 (listening/reading/writing advanced, speaking intermediate)
English
Technical reading and writing; Dev.to articles and English specs