Haishan Tan
AI Product Engineer / AI Product Manager
English (Fluent)Mandarin (Business)Cantonese (Business)
Professional Summary
AI Product Engineer with 10+ years of software experience and a Forward Deployed Engineer working style: enter the business, define where AI creates value, design agent behavior, and ship production systems.
Over the past two years I have led AI-native products from 0→1—product discovery, MVP definition, agent workflow / HITL governance, architecture, and delivery—while partnering closely with operators and enterprise stakeholders.
Previously led front-end engineering for large QSR digital platforms related to McDonald's China and global digital menu operations (10+ markets, 100,000+ devices), making product and technical trade-offs under real operational constraints.
Work Experience
Senior AI Product Engineer / Founding Engineer
- Own AI-native products end-to-end in an FDE style: discovery with real operators, product definition, agent workflow design, architecture, delivery, customer communication, and commercial proposals.
- Lead two products—PopPic (restaurant OS) and SurveyMap (smart mobility survey)—from problem framing through pilot delivery.
- Establish agent runtime, tool contracts, command layer, and human-in-the-loop approval so AI mutations are permissioned, approvable, auditable, and replayable.
- Consolidate delivery from 5 separate repositories into a business-modular monorepo / microservice architecture to improve access performance, reduce infra cost, lower coupling, and keep teams shipping independently.
Project
Founding Product Engineer — PopPic Restaurant AI OS
- Completed discovery across 200+ restaurant sites and 35 operators (coffee, pizza, burger, Western dining); after ~6 months of iteration, pivoted from a generic AI image tool to a restaurant vertical OS.
- Redefined scope to POS + AI Agent + Admin + Mini Program + Marketing, with the goal of turning “launch a new item” into an assisted operating workflow rather than a chatbot demo.
- Designed HITL agent workflow (intent → plan → permission → tool selection → human approval → execute → audit) so AI never becomes the system of record for business writes.
- Replaced prompt-heavy UX with template / style / visual selection and one-click generation; shortened new-item prep & launch from ~3 days to ~2 hours on average in pilot workflows.
- Architected Local-first POS, agent runtime, command layer, and audit/approval systems with React, Rust POS, Node, PostgreSQL, SQLite, OpenAPI, and Docker.
Project
Lead Product Engineer — SurveyMap Smart Mobility Platform
- Owned full product cycle with the institute: planning, PRD, system design, and delivery across project → survey point → field work → review → archive → GIS → AI.
- Designed RBAC for project leads, designers, field surveyors, and owner visitors; rebuilt survey state machine and GIS data model for production survey operations.
- Built AI review / assistant capabilities on the client’s internal DeepSeek + MCP knowledge base with isolation controls so sensitive material does not leak to public cloud; reconciled remote data into map-layer UI for traffic and survey context.
- Delivered with React, NestJS, PostgreSQL, OpenLayers, Three.js, and monorepo architecture; one Haizhu District project alone covers 4,000+ survey points.
Front-End Development Manager
- Led digital platforms related to McDonald’s China operations across CRM, marketing, campaign, coupon, store management, and customer operations.
- Owned front-end architecture for global digital menu / CMS platforms serving QSR brands including McDonald’s and Hungry Jack’s, spanning 10+ markets (US, Canada, Australia, Middle East, Europe, Russia/Ukraine, Hong Kong & Macau) and 100,000+ in-restaurant and outdoor endpoints.
- Migrated core systems from Angular to React + TypeScript; improved loading, payload, API, and CI pathways for markets with 10,000+ devices, lifting combined performance metrics by ~30%.
- Accelerated iteration delivery through tooling and engineering standards so sprint cycles completed ahead of plan; operated under SLA for restaurant-side content faults (wrong content, black/white screen) and scheduled marketing deployments.
Key Achievements
~30% combined performance improvement on large-device markets via load, bundle, API, and CI optimization
Faster iteration throughput through process/tooling upgrades (cycles completed ahead of schedule)
Production operations covering SLA fault response and timed marketing content rollout across restaurant endpoints
Project Manager / Technical Supervisor
- Delivered 10+ corporate site rebuilds with agile process and SEO/performance work that increased organic traffic by ~120%.
Junior Web Developer
- Built multilingual WordPress themes and form plugins; optimized queries and lead-management workflows.
Education
Bachelor of Science
Simon Fraser University | 2013
Mathematics and Statistics
Skills
Keywords
- TypeScript
- React
- Next.js
- Node.js
- NestJS
- Python
- Rust
- PostgreSQL
- Redis
- Docker
- OpenAPI
- OpenLayers
- Three.js
- TurboRepo
- LLM
- Agent
- Tool Calling
- Context Engineering
- HITL
- RBAC
- GIS
AI Product Design
- AI-native product discovery / MVP / positioning
- Agent workflow design & human-in-the-loop governance
- Prompt / context system design & AI evaluation
- Forward-deployed product collaboration with operators
Engineering & Architecture
- TypeScript / React / Next.js / Node.js / NestJS
- Python / Rust / PostgreSQL / Redis / Docker / OpenAPI
- Agent runtime, tool contracts, command layer, event-driven systems
- Local-first architecture, monorepo / microservices, DDD & API design
Product & Leadership
- Business analysis, user research, PRD, roadmap
- Cross-functional collaboration & technical mentoring
- Enterprise stakeholder communication (design institutes / QSR ops)