Tokyo, Japan · Open to opportunities · Available immediately

Senior AI Product Manager

AI Products · Recommendation Systems · Knowledge Graphs

8+ years building content ecosystem, algorithm strategy, and AI products across large-scale consumer platforms including Autohome, Tencent, Bilibili, and Alibaba.

View case studies Contact

Core Tags What I bring into AI product roles

Selected Work

AI and algorithm products with production impact

Autohome · AI Agent · 2025

Automotive AI Q&A Agent

Led the query intent system, evaluation framework, and knowledge architecture rebuild for an automotive AI assistant. The core product judgment was to treat hallucination as a domain-knowledge problem, not only a prompt or model problem.

Agent product design Knowledge architecture Evaluation framework

AI Agent + Knowledge Layer Rebuilt the answer engine around domain knowledge, not only model access

Before

User query DeepSeek answer Hallucination quality 50%+

My product move

Locate the issue in the knowledge layer

intent system retrieval rules evaluation criteria

After

Intent Knowledge retrieval Grounded answer coverage 95%+ · quality 70%+

Three-layer automotive knowledge foundation

Knowledge Graph

Car Series Trim Price Config Competitor

Knowledge Base

Brand Series Model year Spec / policy / review

Curated Content

Expert article Owner review Editorial guide

Sample query scope

SUV under RMB 200K

Trim / price comparison

Failure diagnosis

AI assistant structured answer screenshot

Bilibili · Knowledge Graph

Multi-dimensional Tag Knowledge Graph

Owned the product structure, metrics, and acceptance standards for a content understanding graph that replaced fragmented tag models. The graph helped represent videos across topic, scene, ingredient, style, and intent dimensions.

Content understanding Graph product structure Cold-start recommendation

Before

User tag Many small tag models Coverage stuck around 70%+ Long-tail labels could not scale

Graph Product

After

Dynamic multi-dimensional tag graph

98%+ content coverage 90%+ recall / accuracy

Topic Scene Entity Weight

Autohome · Recommendation Platform

One-stop Recommendation Service Platform

Turned repeated short-cycle recommendation requests into a self-service platform. Operations teams could configure audiences, content pools, modules, and business rules without reopening the same product, algorithm, and engineering workflow each time.

Recommendation ops platform Scenario configuration Delivery efficiency

Business transformation Temporary recommendation scenarios became a self-service launch system

Before

Campaign requestPM / algorithm / engineer coordinationOne-off delivery

My product move

Audience Content pool Recall Ranking Rules Monitoring

After

Self-service configAuto runLaunch & monitor

Outcome

19h → 1h 90%+ less repeated delivery effort

Autohome · Commercial Algorithm

Lead Intent Scoring Model

Reframed the model problem by correcting the training label: phone-stated intent was not the same as real transaction. By using verified downstream transaction signals, the scoring model became useful for lead allocation and delivery planning before delayed conversion feedback returned.

Before / After Lead quality moved from delayed black box to real-time scoring

Before

User leadLead delivery?Dealer / OEM

Had to buy supplemental store-visit, test-drive, or conversion outcomes to protect customer renewal

Quality was invisible at delivery time

After

User leadReal-time scoreHigh / Mid / LowSmart routing

Ground truth rebuilt with verified downstream transaction signal

Predict conversion earlier · allocate better · reduce external buying

Business chain lead → visit → test drive → purchase

Problem found phone intent ≠ real transaction

                  My decision
                  redefine ground truth
                

Product value Turned lead delivery from volume selling into quality-based allocation

Predict earlier Estimate conversion quality before the offline purchase result returns.

Allocate smarter Route high-intent leads to better-fit customers and delivery plans.

Reduce external buying Lower dependency on supplemental store-visit, test-drive, and conversion purchases.

0.91 transaction correlation 9.7x high / low separation USD 700K+ supported renewal impact

Initial application window: Nov 20-Dec 19, 2025. Approx. USD equivalent.

Additional AI Product Work

Beyond the main cases

Tencent News Content Intelligence

Selected content intelligence work across tag-based supply optimization, recommendation quality, content diversity, and company-wide platform risk governance initiatives. Public metrics are kept at 90%+ coverage, accuracy, and recall.

AIGC marketing content platform

Researched and designed an AI workflow covering copy, posters, video, live/digital human concepts, multi-platform publishing, and lead landing pages. Not launched, so it remains an additional design case.

Large-model Content Leads

Explored how LLM-based content understanding and user intent signals could convert high-intent content consumption into automotive lead opportunities. Kept as a reserved case until the project boundary is fully reviewed.

Alibaba Youku Content Taxonomy

Built second-level content classification and ecosystem decision frameworks for a large digital entertainment platform, supporting structured supply understanding and operations decisions.

Platform Risk Governance

Worked on company-wide platform risk governance initiatives across content ecosystems, using public-safe language and abstracted project descriptions for external review.

AI Product Research Portfolio

Additional research and design work across AI marketing, AIGC video, digital human concepts, publishing workflows, and lead landing pages.

Experience

Large-scale consumer platforms, AI, and content ecosystems

2022.08 - 2026.02AutohomeSenior AI Product Manager · AI, recommendation, commercial algorithm products

2021.08 - 2022.07Tencent · Tencent NewsContent intelligence, recommendation strategy, platform risk governance

2020.09 - 2021.08BilibiliContent understanding knowledge graph, ecosystem strategy, platform governance

2017.12 - 2020.09Alibaba Digital Media & Entertainment Group · YoukuContent classification, ecosystem strategy, platform operations

Early experience

2016.08 - 2017.08Fabon Information TechnologyOn-site Product Assistant for Tencent QQ Zone project

Contact

Open to AI product roles in Tokyo and global teams

Best fit: AI Product Manager, Algorithm Product Manager, Recommendation/Search Product, Knowledge Graph/RAG/Agent, and Content Ecosystem roles.

080-8142-1160 hancy_han@163.com LinkedIn