30%
不透明清透

A story about learning to see

Tap to reveal
Ian Weng
Ian Weng 的透明玻璃 Instagram 個人檔案卡
Light, distance, a moment.

Science Researcher & Photographer

Explore selected work
Scroll

00 / Prologue

Every project began with
something worth noticing.

A pitch moves too fast for your eyes to follow. A patient’s hand twitches without meaning to. Health numbers pile up and nobody quite knows how to read them. I keep asking the same thing: can technology help us see what’s hiding in there, and actually use it?

01 / Learning Ways of thinking

I'm curious about a lot of things, so I write code to find patterns, build hardware to make things real, and use a camera to learn how to see.

Development

Python, OpenCV, NumPy, TypeScript, React, Firebase, TensorFlow.js

See PitchMotion

Hardware

Arduino, RFID, EMG signal processing

See the warning bracelet

Design

Interface design, photography

See the photographs

03 / Making Selected work

From sketching a home to reading a fast pitch: explore the question, the system, and the result.

Sketch illustration of a patient wearing a detection bracelet that monitors EMG movement and NG-tube proximity
  1. Quietly monitoring
  2. RFID checks distance
  3. Hand moves, EMG spikes
  4. Bracelet alerts

NG-tube Warning
Bracelet

Warn caregivers before the tube comes out.

EMG-based system detecting involuntary hand movements in NG-tube patients. Integrated RFID sensors into Arduino to monitor proximity and trigger alerts via a soft restraint mechanism, reducing unintended tube-removal incidents in clinical settings.

Arduino, EMG, RFID, mobile app

Problem, build, and outcome
Problem

Unintended tube removal can happen before caregivers have time to react.

Build

Connected EMG sensing, RFID proximity checks, and alert logic into a wearable prototype.

Outcome

Creates an earlier warning layer while keeping the restraint softer and more humane.

Sketch illustration of the PulseBridge health dashboard showing ECG, HRV, sleep, readiness, and AI health analysis
  1. Export watch data
  2. Heart and sleep load in
  3. Metrics become charts
  4. AI read-out for the visit

PulseBridge

Bring your watch data into the clinic.

Upload Apple Watch ECG, HRV, and sleep exports; complete a five-step symptom questionnaire. Get bilingual AI health analysis that bridges Western biomarkers with Traditional Chinese medicine, with follow-up chat and one-click PDF for clinical visits. Next.js frontend, FastAPI backend on Cloud Run.

Next.js, FastAPI, Groq, Apple Health, Firebase

Problem, build, and outcome
Problem

Personal health exports are rich in data but hard to interpret before a clinical conversation.

Build

Designed an intake flow, AI analysis layer, follow-up chat, and PDF handoff for visits.

Outcome

Bridges wearable data and human symptoms into a clearer pre-visit narrative.

Sketch illustration of the CashFlow investment dashboard with portfolio allocation, market watchlist, and Monte Carlo simulation
  1. Set a portfolio
  2. Run Monte Carlo
  3. Compare the ranges
  4. Spot what changed

CashFlow

Turn investment uncertainty into readable ranges.

Full-stack investment cash-flow simulator and market dashboard for beginner investors. Runs Monte Carlo projections, ships portfolio templates and TW / US watchlists, and pairs a Next.js + Recharts / Three.js front end with a FastAPI backend. Optional Finnhub, OpenAI / Groq, and Firebase integrations.

Next.js, FastAPI, Recharts, Three.js

Problem, build, and outcome
Problem

Beginner investors need to compare risk, allocation, and cash-flow futures without drowning in jargon.

Build

Built dashboard views for portfolios, watchlists, simulation ranges, and market context.

Outcome

Turns uncertainty into ranges and scenarios that are easier to discuss and revise.

Sketch illustration for Chi Na Jian: a phone showing a neighborhood map and three nearby restaurant recommendations
  1. Three places nearby
  2. Pins on the map
  3. Pick one
  4. Navigate there

Chi Na Jian
Restaurant Picker

Three nearby picks, each with a reason.

Mobile-first nearby restaurant decision tool that returns three curated options with verifiable reasons. It supports current or manual location, list and map views, opt-in translation, favorites sync, and graceful fallbacks when external services fail.

React, Google Maps, Firebase

Problem, build, and outcome
Problem

Choosing a nearby restaurant often means sorting through too many results and opaque recommendation scores.

Build

Built a guest-first flow that combines location, food intent, three transparent recommendations, and native map navigation.

Outcome

Turns a broad nearby search into a decision people can inspect and act on without creating an account.

04 / Timeline Along the way

No journey is finished alone. From research to competitions, I learned to listen, collaborate, and keep refining ideas until they grow beyond where they began.

Selected recognition

Each stage taught me to explain an idea, defend it, and make it stronger.

  1. International Engineering Science Festival (I-FEST2)

    Gold medal, Tunisia

  2. Taiwan International Science Fair (TISF)

    2nd place, engineering

  3. International Exhibition for Young Inventors (IEYI)

    Silver award