Intro
Hi, I'm Bo An Chen (陳柏安).
I'm a student majoring in Information Management
at National Taiwan University of Science and Technology.
I enjoy using technology to make things simpler and more intuitive.
Whether it's building an app, integrating AI, or designing a system workflow,
what matters most to me is that the people using it find it helpful and natural.
I believe great products come from understanding people's needs — not just
what's technically possible, but what truly feels right to use. That's what I care about most in
every project I take on.
Feel free to look around and learn more about my story.
Work
Here are some of my featured projects. Click on any project to learn
more.
WhatEat APP
AI-Powered Restaurant Recommendation
WhatEat is a university capstone project that solves the "what to eat"
decision paralysis problem. Unlike traditional food apps that rely on keyword searches, WhatEat uses
AI to proactively recommend restaurants based on your implicit needs.
Key Features
- Contextual Understanding: Uses LLM to understand
implicit needs (mood, weather, dining purpose)
- Explainable AI: Tells users WHY a restaurant is
recommended, building trust
- Smart Search: AI-powered conversational search
interface
- Personalization: Dietary preferences and
restrictions support
System Architecture
A hybrid RAG (Retrieval-Augmented Generation) architecture
combining SQL Server and Vector Database.
- Intent Analysis: LLM parses natural language
into structural JSON constraints.
- Retrieval: Fetches candidates via SQL (User
Data) and Google Places API.
- Personalization: Re-ranks results based on
historical preferences and weighted features.
- Generation: LLM synthesizes the final answer
with reasoning based on retrieved data.
Technical Highlights
- Anti-Hallucination: Strictly constrained RAG
allows the AI to only recommend real, existing restaurants.
- Hybrid Analysis: Merges Unstructured Data
(Reviews) with Structured Statistics for precise personalization.
- Dynamic Context: Adapts to GPS location, time
of day (Lunch/Dinner), and current weather.
Screenshots
Conversational Search
Visual Map Selection
AI Recommendations
Reasoning & Details
Rating & Feedback
Personal Preferences
Tech Stack
Android (Kotlin), LLM API, Google Places API, SQL Database
Team
BO-AN CHEN (陳柏安) - Full Stack Development
Yi-Hao Dong (董亦浩) - @kivxxx
KM_CSS
Customer Satisfaction Survey & Service Management System
KM_CSS is an enterprise-grade customer satisfaction survey and service
management system developed during my internship at KENMEC. The system enables end-to-end management
of customer satisfaction surveys, service tracking, engineering progress monitoring, and data-driven
analytics — all through a modern dark-themed admin dashboard.
Key Features
- Analytics Dashboard: Real-time overview with KPI
cards, Chart.js visualizations (bar charts, doughnut charts), and latest feedback display
- Survey Management: Token-based survey link
generation, customizable survey questions, and customer satisfaction data collection
- Engineering Progress: Track maintenance and
renovation projects with ERP data synchronization, quotation tracking, and status management
- Customer & Employee Management: Full CRUD
operations for customer and employee records, employee-customer mapping, and shift scheduling
- Satisfaction Analysis: Per-customer satisfaction
trends, score distribution analysis, service type statistics, and time-filtered detailed reports
- Service Request Portal: Customer-facing forms for
submitting service requests with contact info, problem descriptions, and photo uploads
- Employee KPI: Automated KPI calculation based on
customer satisfaction scores linked to assigned technicians
- AI Service Report Analysis: Integrated LLM-powered analysis that automatically grades service request severity (Emergency / Medium / Normal) and generates concise AI summaries for rapid triage
Security Implementation
- Multi-level RBAC: Session-based authentication
with 3-tier permission levels — Admin (Level 2), Department Member (Level 1), and Guest (Level
0)
- Token-based Access: Unique cryptographic tokens
for each customer survey link with expiration and activation controls
- Input Validation: Server-side strict validation on
all API endpoints to prevent SQL Injection and XSS attacks
Tech Stack
Flask (Python), SQLite, Vue.js 3, Element Plus, Chart.js, Groq AI (LLM)
My Role
Full Stack Developer (Internship Project)
Screenshots
Admin Dashboard
Engineering Progress
Customer Statistics
Satisfaction Analysis
Service Statistics
AI Service Report Analysis
Admin Login
TX-Observer
Automated Taiwan Futures & Spot K-line Chart Push Notification System
TX-Observer is a fully automated technical analysis push notification bot deployed on a headless Linux server. The system follows strict trading session schedules to automatically fetch market data via Shioaji API, generate professional-grade K-line charts with moving averages, and push them to Discord and Telegram — all without any GUI or manual intervention.
Key Features
- Dual-Period Combined Charts: Futures: 60K + 5K dual-panel chart; Spot: Daily K + 60K dual-panel chart, with MA5/10/20/60/240, high/low annotations, and Doji candlestick highlighting
- Smart Scheduling: Dual-layer protection — XTAI market calendar (handles holidays, Chinese New Year, typhoon closures, Saturday make-up sessions) + trading session time-gate filter
- Settlement Day Alerts: Automatically detects the third Wednesday of each month (futures settlement day) and prepends a settlement day notice to push messages
- Data Freshness Diagnostics: Three-layer guard (session / holiday / Snapshot comparison) with automatic token-expiry re-login and stale data detection
- Dual-Platform Push: Discord Webhook + Telegram Bot with independent error isolation — one platform's failure never blocks the other
- Error Isolation: Single-symbol failures are logged without affecting other symbols; the scheduler keeps running without crashes
System Architecture
Fully automated data pipeline running on a headless Linux VPS:
- Data Layer: Shioaji API → 1-min OHLCV bars → session-aware resample to 5K / 60K / Daily
- Rendering Layer: matplotlib GridSpec single-figure combined chart with mplfinance external-axes mode, dark theme, CJK font support
- Push Layer: Direct binary upload to Discord Webhook (multipart) + Telegram Bot API (sendPhoto), no external image host needed
- Scheduling Layer: APScheduler cron triggers with market calendar integration, closing summary retry×3, and misfire grace handling
Tech Stack
Python, Shioaji API, pandas, matplotlib / mplfinance, APScheduler, Discord Webhook, Telegram Bot API, Linux systemd
My Role
Solo Developer — Full pipeline from data fetching, resample logic, dark-themed chart rendering, to dual-platform push and systemd deployment
Screenshots
60K + 5K Combined Chart
Discord Push Notification
Telegram Push Notification
Links
Experience
2026
TX-Observer — Personal Project
Built a fully automated Taiwan futures & spot index K-line chart push notification system, deployed on a headless Linux server with dual-platform push to Discord & Telegram.
Solo Developer
2025
KM_CSS — KENMEC Internship
Developed an enterprise-grade customer
satisfaction survey & service management system. Built the full stack with Flask, Vue.js 3,
and SQLite.
Full Stack Developer
2025
WhatEat APP — Capstone Project
Led frontend development of an AI-powered
restaurant recommendation app using Flutter and Gemini API.
Team Lead / Frontend
2022 – 2026
National Taiwan University of Science and
Technology
B.S. in Information Management.
Education
About
Skills
- Programming: Python, Kotlin, Dart,
JavaScript,
SQL
- Frameworks & Tools: Flutter, Vue, Android
Development,
Git, VS Code
- Databases: MySQL, SQLite
- Other: RESTful API, LLM Integration
Education
National Taiwan University of Science and
Technology
B.S. in Information Management (Class of 2026)
Relevant Coursework: Database Management, Software
Engineering, Capstone Project
Interests
- Travel and exploring new places
- Researching tools and automation
- Gaming, Anime, Fitness, Web3
Goals
- Short-term: Complete my degree and gain hands-on industry
experience
- Long-term: Deepen expertise in Automation, Quantitative Trading, and
Web3
Personal Traits
- Detail-oriented with a passion for efficiency
- Enjoy solving complex problems with elegant solutions
- Skilled communicator who bridges technical and non-technical
perspectives
- Strong team player with experience in collaborative projects
Contact
