FL
Flutter App Delivery
Business screens, state management, API integration, media upload, edge states, and Android/iOS/OHOS release adaptation.
FlutterDartRiverpod
Jacky Chou
12-Yr Mobile Architect / Flutter Tri-Platform / Full-Stack AI Agent Engineer
2015-now / Mobile / Backend / AI
Shipping tri-platform commercial apps, Java backends, and controlled AI Agent systems end-to-end
With 12 years of Android native and mobile engineering experience, I have delivered two commercial brands across Android, iOS, and OpenHarmony to app stores, and independently ship full-stack workflows from UI design and Java 21 backends to RAG/MCP and multimodal write-back agents.
I focus on high-assurance, verifiable delivery: production mobile apps, native runtimes, RBAC backend APIs, controlled AI boundaries, pre-registered evaluations, and real-device proof.
12 yrs
Android native foundation & commercial app delivery
2 Brands × 3 OS
Shipped across Android, iOS, and OpenHarmony stores
AI Full-Stack
UI → Java 21 Backend → App → RAG / MCP / Write-Back Agent
Capability Matrix
Mobile, backend, and AI delivery
A practical stack for shipping client-ready products: mobile experience, native capability integration, backend systems, AI boundaries, and delivery evidence.
AN
Android Native Capabilities
Permissions, camera, location, QR scanning, push, sharing, WebView/MethodChannel, Gradle builds, and real-device debugging.
JavaKotlinGradle
AI
AI Engineering
RAG, embeddings, pgvector, OpenAI-compatible providers, local models, MCP tools, Agent workflows, and human handoff boundaries.
RAGMCPAgent
Selected Work
Case studies with delivery evidence
The selected cases emphasize responsibility, architecture, AI control boundaries, native app capabilities, and validation signals.
Primary Flutter Case
Flutter / Android / iOS / OpenHarmony
Pocket Home Flutter (2 Brands × 3 Platforms)
Delivered two commercial real-estate brands across Android, iOS, and OpenHarmony via Flutter + native integration, all published on mainstream app stores.
FlutterAndroidiOSOHOS
2 commercial brands shipped across Android, iOS & OHOS stores
Mainstream B2B app & IM capability coverage
Native Android/iOS/OHOS capability integration
AI Workflow
Flutter + Vue + Spring Boot + RAG
AI Customer Service Ticket System
An AI-first support workflow where knowledge-base answers are attempted first and low-confidence or out-of-scope questions move into human tickets.
RAGSpring BootVue
Controlled AI response boundaries
Human handoff and audit trail
Reusable MVP pattern for small businesses
AI Data Product
Flutter + Spring Boot + LLM
AI Weekly Business Report
A data-to-report workflow that uploads CSV/Excel files, computes metrics asynchronously, and generates reviewable Chinese business reports.
Async JobLLMDocker
Backend-owned metric calculation
Reviewable AI narrative generation
OpenAI-compatible and local model validation
Knowledge AI
Java 21 / Spring Boot + Python LangGraph + MCP
Enterprise RAG + LangGraph Agent + MCP Server
SQL-level ACL retrieval, backend citation verification, LangGraph multi-step decomposition (6/12 → 12/12), revocation enforcement (0/3 leakage), and official MCP SDK server.
RAGLangGraphMCPpgvector
0/3 post-revocation leakage & 12/12 cross-doc accuracy
Backend-verified citations & fail-closed handoff
182 Java tests + standalone Python MCP server
Multimodal Write-Back Agent
Python + Dual-VLM + Spring Boot Write-Back
Field Document Intake & Idempotent Write-Back Agent
Extracts repair, maintenance, and acceptance forms (PDF, DOCX, scans, phone photos) with high-res second-look validation, deterministic table rules, and idempotent work-order write-back.
Multimodal AgentIdempotencyHITLEval
0 wrong auto-writes on 20 held-out docs & 20 real photos
0 duplicate or lost work orders under fault injection
Pre-registered evaluation discipline + 217 automated tests
Java Full-stack
Spring Boot + Vue Admin
Field Service Ops Platform
A reusable operations platform base covering authentication, RBAC, audit trails, customers, tickets, approvals, and smoke validation.
Spring BootVueRBAC
Reusable backend starter pattern
Admin workflow and API integration
Docker and smoke-test evidence
Proof Evidence
Designed for reviewers, clients, and interviews
The portfolio is structured around evidence: delivery scope, architecture, validation, AI boundaries, and communication-ready project narratives.
Evidence-first delivery
Case studies include screenshots or productized visuals, architecture nodes, validation evidence, and role boundaries.
Controlled AI behavior
AI cases emphasize source citations, out-of-scope handling, human handoff, and replaceable model providers.
Client-ready stack
Mobile apps, Java backends, Vue admin systems, RAG, Agent workflows, deployment, and verification are presented as one delivery system.
Private repository review
Core case-study repositories are kept private to protect project context, but code, commit history, build logs, and runtime evidence can be shared through authorized review when needed.
Contact