01 / UNDERSTAND
Content & human context
Assessing nuance in content, creator fit and audiences — not just numeric performance.
A personal portfolio · Ho Chi Minh City
I didn't start with AI.
I started with a problem.
I'm Amelia Ph. Nguyen — a Business Administration student and early-career operator working across creator operations, commerce and analytics. I care about making complex work clearer, more useful and easier to trust.
Business Administration · UEHData Analytics · VNU-HCM
OPERATIONS × DATA × AI CURIOSITY
Amelia Ph. NguyenFrom business operations to quality systems
Beyond the resume / About Amelia
Business roots.
A systems mindset.
My path into AI didn’t begin in a lab. It began in real-world operations: supporting product launches, reviewing creator content, analyzing customer data and building tools that made everyday work easier.
Across very different teams, I kept coming back to the same question: how can we turn individual judgment into a reliable, repeatable process — without losing the human context?
Explore the journey02 / QUESTION
Data & structured evidence
Separating what is observed from what is assumed, missing or needs validation.
03 / BUILD
Systems & better decisions
Prototyping workflows that reduce repetition and make human review easier.
The path behind the project
From operations to
AI evaluation.
Four chapters, one evolving curiosity. Select a chapter to see what I worked on and what it taught me.
Execution taught me to think commercially.
My first chapter brought me into digital commerce, creator partnerships and product launches. Co-leading TikTok Shop livestreams taught me to connect shopper insights to concrete decisions and measurable results.
VND 3B GMV2 co-led livestreams · 120% of target
What made me curious about AI ↗Evidence from the work
Measured experience.
Real operational impact.
Results from roles at INNO, 4AM and Mercedes-Benz Vietnam, presented separately from the SIGNAL prototype.
20K+Creator profiles screened4AM · campaign shortlisting
60%Less reporting time4AM · 5h → 2h per ~30 creators
372K+Aftersales records analyzedMercedes-Benz Vietnam · 16 dealers
VND 3BLivestream GMVINNO · 2 co-led sessions · 120% target
Professional experience metrics are separate from synthetic test results in the SIGNAL project.
The turning point
I wanted to spend less time moving information — and more time making thoughtful decisions.
Screening profiles, checking content guidelines and updating reports showed me how much work followed repeatable patterns. I began experimenting with AI-assisted workflows. Then a better question emerged: if a system makes a decision faster, how do we know when it is wrong?
See what that question became ↗THE FIRST QUESTIONWhat parts of this workflow are repetitive?
THE SECOND QUESTIONWhich judgments can be translated into explicit criteria?
THE QUESTION THAT BECAME SIGNALHow do we investigate the decisions that are wrong?
Where curiosity became a tool
Meet SIGNAL.
Inspired by a question from everyday operations: when automation makes a judgment, how can we check its evidence and improve the next decision?
INDEPENDENT PROTOTYPE / 2026
SIGNAL.
An evidence-first evaluation workflow combining creator and content inputs, rule-based rubrics, optional AI text assessment, human review and rubric comparisons.
5 synthetic cases4/5 → 5/5 reference-label matches
The five-case exercise illustrates rubric revision, not real-world model accuracy or proof of generalization.
Enter the SIGNAL case study ↗
ACTUAL PROTOTYPE SCREENSHOTLet's connect
I like people who ask better questions.
I'm interested in thoughtful technology, operations, data quality and the human decisions behind reliable systems. Happy to connect or exchange ideas.
© 2026 AMELIA PH. NGUYENHO CHI MINH CITY · VIETNAMBUILT WITH CURIOSITY