Back to Experience
POI
POI-TECH logo

Mar 2019 — Oct 2024 · Full-time

Product Manager (TL)

Guangzhou POI-TECH Intelligent Information Technology Co., Ltd.

"Building big data products from scratch taught me that the best features are born out of real customer friction. Over five years, leading 3 PMs and collaborating across 100+ sprints to scale our SaaS and AI portfolio wasn't just about shipping code — it was about bridging business vision with deep engineering execution."

Product Scope

Product Ecosystem

Owned product strategy across POI-TECH's industrial data ecosystem, connecting data integration, metric management, visualization, AI applications, AIGC services, and MOM platform modules into one scalable enterprise product portfolio.

POI-TECH product ecosystem diagram

Product Contributions

SQLAPIMySQL

Data Integration Tool

Data Integration Tool illustration

Enterprises needed to solve data integration across multiple systems while reducing connection costs. Metric configuration and visualization were also blocked by scattered data cleaning, exporting, and reporting workflows.

Design and deploy a configurable data service platform covering metric management, data integration, and OpenAPI access for heterogeneous internal and external systems.

Metric LibraryDesigned Basic Metrics, Derived Metrics, and Interface Metrics to prepare calculated indicators for downstream BI and analysis scenarios.
Data IntegrationBuilt configurable data collection, data table, query, export, and compute-task modules with MySQL backend, enabling flexible ingestion from IoT, MOM+, QMS, ERP, MES, and WMS sources.
OpenAPIStandardized business data access through unified Create/Update and query APIs, plus API documentation and request/response parameter management.

Served 20 enterprises with 500 API integrations in 1 year. Reduced single-interface delivery time from 16 hours to 4 hours or less.

  • Supported 80% IoT/sensor collection, 10% internal system data, and 8% external system data integration patterns
  • Launched 20+ internal system OpenAPIs
  • Covered 90+ indicator calculation scenarios
HTMLJavaScriptD3.js

Digital Factory

No-code Data Visualization Product

Digital Factory illustration

Manufacturing teams needed personalized data visualization and enterprise-customized dashboards without relying on repeated development support.

Lead the design of a configurable Digital Factory dashboard product that lets teams assemble dashboards quickly through reusable components, templates, and real-time data integration.

Visualization ComponentsDesigned a component system covering charts, data cards, tables, and configurable display widgets for common industrial monitoring scenarios.
Real-time IntegrationDefined live updates, component linkage, drill-down, filters, data source management, metric library, and API integration workflows.
Delivery TemplatesBuilt template management, permission control, and multi-platform preview capabilities to accelerate customer-specific dashboard delivery.

Shipped a reusable no-code dashboard platform that lowered customized dashboard delivery cost and supported enterprise-scale factory visualization.

  • Dashboard inventory grew to 3,000+
  • Completed low-code digital transformation in 1 month
  • Reduced Single Digital Factory delivery time from 3 months to 1 month
HTMLJavaScriptD3.js

Reporting Platform

No-code Report Design & Export Tool

Traditional offline report creation was slow, repetitive, and dependent on manual monthly work, creating a bottleneck for manufacturing analysis.

Digitize offline report workflows and improve report creation efficiency for daily and monthly reports.

Report Type SystemDesigned report templates for classical reports, traditional Excel-style reports, and business reports.
Report DesignDefined data selection, condition editing, preview, sharing, and export workflows for non-technical users.
Access ControlAdded permission and access-control patterns to support enterprise report management across departments.

Enabled low-cost, self-serve report creation and export for manufacturing users, reducing manual reporting workload at enterprise scale.

  • Reached 80%+ report scenario coverage
  • Used by 500+ enterprises
  • Reduced repetitive monthly report work through reusable templates
Linear RegressionMMOEXGBoost

Industrial AI Operations Research Application

Industrial AI Operations Research Application illustration

Industrial operations heavily relied on human expertise that was difficult to inherit and replicate. To maintain stable product quality in continuous production, manufacturers needed intelligent operations management.

Lead business research and AI product planning for real-time quality prediction and process optimization, connecting scenario mining, evaluation metrics, algorithm models, and product iteration.

Value Scenario MiningIdentified high-value production scenarios and translated operational pain points into measurable model goals and product requirements.
Model CoordinationCoordinated AI engineers across model building, training, and optimization using Linear Regression, XGBoost, and MMOE approaches.
Product IterationConnected raw-data OpenAPI inputs and production feedback loops to continuously refine model metrics, recommendations, and monitoring views.

Deployed across 23 production lines in 5 enterprises, reducing costs by $3M annually. Recognized as an outstanding case of AI-powered industrial innovation at 2024 WAIC (World Artificial Intelligence Conference).

  • 90%+ real-time prediction accuracy
  • WAIC-recognized outstanding case
  • Reduced energy cost by $1M+ in one optimization scenario
RAGSupervised Fine-Tuning

AIGC Application Service

AIGC was gaining momentum, but MVP exploration required a specialized industrial corpus combining metric data, paper-industry knowledge, and private enterprise data.

Lead the product team to research, plan, and ship AIGC-powered services for natural language operational analysis and industrial knowledge retrieval.

Market ResearchLed product team in market research of 6+ mainstream LLMs and completed end-to-end AIGC product planning and design.
Knowledge BaseBuilt a knowledge base with proprietary data, documents, IoT data, and metrics using RAG, vector database retrieval, and prompt orchestration.
Launch & AdoptionCoordinated product development, supervised fine-tuning, and user-facing AI analysis workflows across mobile and web scenarios.

Acquired 10+ customers for production training within 3 months of launch, with 20+ potential paying clients in the pipeline.

  • Calibrated by professionals and cross-checked using another LLM
  • Generated 20+ enterprise expressions of interest
  • Launched MVP in 3 months
IoTAPSMESEAMWMSBig Data

Process Cloud MOM+ 3.0-4.0

Industrial operations lacked an integrated platform to manage diverse factory systems — from IoT data to ERP modules — resulting in data silos and operational inefficiencies.

Collaborate on building a comprehensive industrial MOM platform with 8+ modules, supporting high-performance data integration at enterprise scale.

Platform DevelopmentCollaborated with the product team to develop an industrial platform with 8+ modules, including IoT, APS, MES, EAM, and WMS.
Big Data IntegrationCoordinated with big data engineers to clean and integrate 100 TB/yr+ of data, enabling high-performance and real-time queries.

Platform currently serves 400+ enterprises with 2,000+ daily active users.

  • Integrated 100 TB/yr+ enterprise data
  • Serves 400+ enterprises
  • Supports 2,000+ DAU