劉 俐 伶
AI & Automation Engineer | Senior Technical Artist
Building AI tools, automation workflows, technical pipelines, and
real-time production solutions across Art, Engineering, and AI.
Internal Project Operations & Automation System
Employee-aware reminders, project file automation, device tracking, ERP integration, and AI-assisted document organization.
AI & Automation Engineer
Designed and developed an internal operations platform integrating employee-aware reminders, project-server synchronization, AI-assisted document classification, controlled NAS filing, device management, ERP work-log workflows, and cross-platform desktop deployment.
#Workflow Automation · #Internal Tools · #AI Classification · #ERP Integration · #Cross-platform
Portfolio Reconstruction · Fictional data · Proprietary details omitted
01 — Overview
Internal Project Operations & Automation System is an internal workflow platform designed to connect employee identity, ERP work-log reminders, project-server synchronization, AI-assisted file classification, controlled project storage, device tracking, and admin monitoring into a unified operational workflow.
The production system was implemented as a cross-platform Electron desktop application with a Python backend service. It integrated employee data, ERP work-log status, project storage, notification logic, device binding, and AI-assisted document processing to reduce repetitive administrative work and improve project-file organization.
Portfolio Reconstruction
This web demo recreates the original production workflow using fictional employees, projects, devices, schedules, documents, and operational data. Production ERP, NAS, employee databases, project servers, and internal APIs have been omitted or disabled.
02 — Problem
Fragmented Internal Operations
Employee reminders, project files, device ownership, ERP work logs, and project storage were handled across separate systems and manual steps.
Missed Work-log Submissions
Employees had different schedules, so fixed-time reminders were ineffective and often generated unnecessary notifications.
Unstructured Project Filing
Project files needed to follow department-specific folder structures, but manual filing created inconsistency and extra administrative work.
Project Discovery
New projects created on the central project server needed to become visible to users without requiring manual project setup on each workstation.
Employee & Device Lifecycle
New hires, resignations, device ownership, application versions, and stale devices required centralized visibility.
File Discoverability
Even after files were stored, users still needed a way to search documents by filename, summary, project requirement, or meeting decision.
03 — My Role
Role: AI & Automation / Full-stack / Internal Tools / Technical Solution Development
I worked across workflow planning, Electron desktop development, Python backend services, employee and device binding, ERP integration, project-server synchronization, AI-assisted file classification, NAS/project-storage workflows, notification logic, admin monitoring, and Windows/macOS packaging and deployment.
Product / Solution
-
Workflow & Requirement Analysis
-
Internal Tool Planning
-
ERP Reminder Logic
-
Project Storage Workflow
-
Department Template Design
-
Technical Feasibility
-
Admin Workflow Design
Engineering
-
Electron Desktop Development
-
Python Backend / API Service
-
AI File Classification
-
Employee / Device Binding
-
Project Server Integration
-
Notification Scheduling
-
App Version Monitoring
-
Windows / macOS Packaging
Designed the system around existing enterprise data sources rather than duplicating employee schedules, project information, or operational records locally.
04 — Platform Capabilities
Employee-aware Work-log Reminders
Uses employee work schedules and ERP submission status to determine whether and when reminders should be sent.
Project Server Synchronization
Detects available projects from the central project source so users do not need to recreate project records manually.
AI-assisted File Classification
Analyzes uploaded documents, classifies their type, generates a short summary, and maps them to predefined project folders.
Controlled Project Filing
Combines AI classification with department-specific folder templates so files are automatically placed into valid, standardized destinations.
Employee & Device Management
Tracks employee-device bindings, operating system, application version, last-online status, new hires, and resigned employee devices.
Project Search & Upload History
Provides searchable project-document summaries and upload history to improve traceability and file discovery.
05 — Project File Automation Workflow
The project assistant was designed to reduce manual file organization while keeping folder structure under controlled production rules.
Users select the project and department, upload files or folders, and let the system analyze the document type and content. AI determines the classification and summary, while the final destination always comes from predefined department templates rather than arbitrary AI-generated folders.
Workflow
Project Server
→ Select Project
→ Select Department
→ Choose File / Folder
→ AI Content Analysis
→ Document Classification
→ Department Template Rules
→ Destination Mapping
→ Short Summary
→ Project Storage / NAS
→ Upload History / Search
This hybrid design uses AI for interpretation while keeping storage structure deterministic and maintainable.
06 — System Architecture
The production system separated the desktop client, backend services, enterprise integrations, AI processing, and admin monitoring so each responsibility could evolve independently.
07 — Technical Decisions & Trade-offs
Fixed Reminder vs. Employee-aware Reminder
Decision: Use employee-specific schedules and ERP work-log status.
Why: Different working hours make fixed reminder times inaccurate and noisy.
Duplicate Data vs. Enterprise Source-of-Truth
Decision: Read employee and project information from existing enterprise sources.
Why: Avoids maintaining duplicated schedules, employment states, and project records.
AI-only Filing vs. Controlled Folder Templates
Decision: Let AI classify the document, but restrict destination folders to predefined department templates.
Why: AI provides flexibility in understanding content while deterministic templates preserve production consistency.
Local-only Tool vs. Central Project Synchronization
Decision: Synchronize project availability from the central project server.
Why: Users automatically gain access to new projects without manually recreating them on each workstation.
Generic User Login vs. Employee-device Binding
Decision: Bind a desktop device to a known employee identity during initial setup.
Why: Enables schedule-aware notifications, device ownership tracking, and persistent workstation context.
Manual Rollout vs. Version Monitoring
Decision: Track installed application versions and last-online state centrally.
Why: Makes cross-platform deployment and outdated-client detection easier to manage.
08 — Impact
Reduced Administrative Overhead
Automated work-log reminders, project discovery, file classification, and device monitoring reduced repeated manual checks.
More Consistent Project Storage
Department templates and controlled destination mapping standardized how files were organized across projects.
Smarter Notifications
Reminders were triggered from actual employee schedules and ERP status instead of fixed, universal times.
Improved File Discoverability
AI summaries, classification, project search, and upload history made stored documents easier to identify later.
Better Operational Visibility
Centralized employee status, device ownership, client version, and system health information for administrative monitoring.
The project demonstrates how AI can be embedded into a broader internal operations system—supporting document understanding, workflow automation, enterprise data integration, and cross-platform production tooling rather than functioning as a standalone AI feature.