Frame the real problem
Translate loose needs into decisions, user journeys, and testable requirements without losing sight of the people behind the metrics.
I’m Chenze Zeng, an aspiring AI product manager. I turn ambiguous user and data problems into practical products—from research and data pipelines to launch and iteration.
I am drawn to AI, B2B, and data products—places where clearer structures and feedback loops make decisions easier.
Translate loose needs into decisions, user journeys, and testable requirements without losing sight of the people behind the metrics.
Work from collection and cleaning through labeling, quality control, and analysis to uncover patterns users cannot yet see.
Move from the first prototype to UAT, launch materials, and the feedback needed for the next iteration.
Build with RAG, automation, agent workflows, and APIs while remaining accountable for the product decisions around them.
Each case began with a concrete gap in information, workflow, or user understanding.
University employment records were available only as unstructured lists, making destination patterns hard to understand. I independently took the product from 0→1: structured 20,000+ records across 34 schools, conducted 12 qualitative interviews, made six key data-design decisions, and led the initial launch. Three on-campus sessions produced a 43% event-to-visit conversion; the site reached 203 UV in its first month, and 58% of surveyed users said it helped calibrate their understanding of career outcomes.
Open the web product ↗Led an eight-person team that labeled data for an NFC-card, mini-program, and spaced-repetition learning system. Delivered 3,500+ labeled words with 96%+ dataset accuracy and a 240% productivity improvement, then transitioned into a product management role and wrote PRD v2.0.
Designed and independently delivered an automated workflow connecting Feishu Bitable with the Fadada e-signature API, including status polling and result write-back. The complete flow passed UAT before handover.
Built a retrieval system over course materials, public writing, and ebooks—from collection and cleaning to structured ingestion and retrieval. The index now covers 3,565+ pieces of content.
A selection from an ongoing weekly reflection practice—projects, experiments, and the moments that changed my direction.
Entered Chongqing University through its broad-based Engineering and Energy track, later specializing in Industrial Engineering, and started a self-training system around reading, communication, and consistent weekly reviews.
Joined the AI vocabulary project as Data Lead and led the team’s data-labeling work before transitioning into product management.
Supported a job fair serving 1,000+ employers and 20,000+ graduates. A Python automation reduced ledger-entry workload by 60%.
Conducted research, structured the data, designed the experience, and launched a product that made graduate outcomes easier for students to understand.
Worked on a dataset platform from 0→1 and delivered an HR contract-signing automation that completed a live, paid e-signature transaction.
Preparing two end-to-end portfolio cases while looking for opportunities in AI, B2B, and data product management.
Agent workflows · RAG · Coze · RPA · Feishu API automation
PRDs · prototyping · competitive analysis · user journeys · qualitative interviews
Python · SQL · scraping · cleaning · labeling & quality control · Excel
Vercel · GitHub · Feishu Bitable · Fadada API · Obsidian
I write to explain how I think about products, growth, work, and the systems behind them.
I write about product thinking and growth, including “Good Products and Bad Products: Short-term Pleasure or Long-term Options?” and a self-developed 3L model for evaluating large technology companies.
Since my first week of university, I have maintained structured reviews of what I did, what I learned, and which judgments turned out to be wrong.
An interview podcast examining choices, costs, and outcomes. Every conversation asks: what variable changed, what did it cost, and what lessons can others apply?
I keep learning across cognitive science, systems thinking, product practice, and long-term decision-making, then test those ideas against real projects.
AI · B2B · Data Products
Shenzhen · Hangzhou · Guangzhou · Chengdu