AI-native product builder · Sydney / Shanghai

Give complexity an orbit.

I turn messy problems into AI products people can actually use.

I frame user problems, design product workflows, orchestrate models and tools, and own the decisions from prototype to public beta. My strength is translating emerging AI capability into a clear, testable product experience.

Selected work

From scattered learning friction to one usable product system.

AI Study Mate began with an ordinary but costly problem: students were surrounded by information, yet still lacked a coherent way to capture, understand and revisit it. I treated that friction as a product-design problem before treating it as a technology problem.

01 · Flagship product

AI Study Mate

A multimodal learning workspace that turns lectures and fragmented materials into structured notes, source-aware answers and review workflows.

I defined the user problems, product flow and acceptance criteria, then coordinated models, retrieval, tools and coding agents to deliver the product. I test outputs, investigate failures and decide what ships, what rolls back and what gets redesigned. The product is now in public beta and operated through an Australian sole-trader entity.

  • Product definition
  • AI-assisted delivery
  • Model & tool orchestration
  • Acceptance testing
  • Public-beta iteration
One learning loop · four connected moments
01 / IN
Capture Live transcription, course files and fragmented learning materials
02 / MAP
Structure Convert raw information into usable notes and navigable concepts
03 / FIND
Retrieve Ask across sources while keeping answers connected to evidence
04 / LOOP
Review Turn understanding gaps into a repeatable revision path
Product decision

Design the learning loop, not a catalogue of AI features.

I prioritised the path from capture to review, so each capability answers a real learning moment instead of becoming another disconnected tool.

My ownership

Keep one accountable product thread from idea to release.

I own problem framing, requirements, interaction paths, task decomposition, validation, release acceptance and iteration priorities—including VPS, domain, deployment and operational decisions.

Working model

Use AI agents as an implementation system, with human judgement in charge.

Coding agents propose and implement within the context and constraints I provide. I compare behaviour against acceptance criteria, trace failures and make the final product and release calls.

How I build

A disciplined loop for AI-assisted product delivery.

I do not position myself as a traditional software engineer. My advantage is designing the working system around the technology: clear context, constraints, hand-offs, verification and decisions that turn uncertain capability into a dependable product.

01

Frame

Find the real user pressure, desired behaviour and evidence of success.

02

Design

Shape the product flow, boundaries, states and acceptance criteria.

03

Orchestrate

Give models, tools and coding agents the right context, tasks and constraints.

04

Verify

Test the full experience, inspect failure modes and challenge plausible outputs.

05

Iterate

Prioritise fixes, release deliberately and learn from real user behaviour.

I own

  • Problem framing, product direction and experience design
  • Requirements, task decomposition and model/tool choices
  • Acceptance tests, failure analysis and iteration priorities
  • Release decisions, operational oversight and user feedback loops

AI and coding agents assist

  • Implementation drafts, scaffolding and repetitive development work
  • Alternative approaches, debugging hypotheses and technical research
  • Code-level execution across frontend, backend and infrastructure tasks
  • Documentation and checks that I review against product intent

Transfer & experiments

The same product discipline, tested in different contexts.

I look for evidence that an approach can travel—from learning to culture, enterprise information and operational workflows—without pretending every prototype is a finished product.

Product transfer · 2026

Guo Ge Hao Nian

Adapted AI Study Mate’s underlying capabilities into a Spring Festival cultural experience, reshaping the flow and presentation for a new audience and use case.

Shanghai City Best Popularity Award · Tangquan Hackathon S1
Team proposal · 2026

Feishu AI Pioneer Future Talent

For a NavInfo enterprise-data Agent proposal, I lead product direction, system framing, evaluation design and presentation integration, transferring relevant lessons from AI Study Mate while keeping the team prototype clearly distinct.

Enterprise information workflow · collaborative proposal
Business workflow · 2025–26

AIA operational automation

Mapped a repetitive channel-operations task and delivered an AI-assisted automation workflow, refining the output through stakeholder feedback and practical business constraints.

Internship application · stakeholder-tested delivery

Background

Product instincts shaped by business, data and communication.

Feb 2024 — present

The University of Sydney

Bachelor of Science

Financial Mathematics & Statistics · Business Information Systems

2025–26

AIA Insurance · Chinese-Funded Strategic Channel Intern

Supported channel operations and client-data workflows; translated a repetitive task into an AI-assisted automation workflow and refined it with stakeholder feedback.

2024–25

Cathay Asset Management · Operations Management Intern

Contributed to VAT-policy research, operational-report accuracy and cross-border collaboration, connecting regulatory detail with business context.

2023

Shanghai TV Finance Channel · Video Editing Intern

Produced short-form financial news under broadcast deadlines and built a searchable video-asset library to improve newsroom retrieval.