HydrAI-SWE

Intelligent Snow Water Equivalent Prediction System

Home Dashboard Analysis Training Knowledge Base Hydropower Agriculture About
Xiao Li

Xiao Li

Senior AI Product Leader

With 20+ years of experience bridging high-level commercial strategy with hands-on technical execution, I possess a unique career arc: from monetizing data for major ad networks to founding an AI/ML SaaS platform, and now, developing production-grade Deep Learning systems.

  • Business Leadership Leading teams, building data ecosystems, and holding P&L responsibility for 80M+ CNY revenue streams in mobile, ad-tech, and VR.
  • Entrepreneurial Execution Independently designed, built, and launched a complete data-driven SaaS platform, leading its technical pivot to modern LLM-based AI.
  • Technical Upskilling Advanced AI Diploma at PACE, UWinnipeg. Expertise in Deep Learning (GRU, LSTM) and full-stack AI (Python, FastAPI, Docker).

Seeking a senior role in AI Product Management or AI Solutions Architecture in Canada to deliver significant business impact.

Connect on LinkedIn Test-AI SASS

🚀 Project Milestones

December 2025
Global Reach & Intelligent RAG
  • Multilingual Vector RAG: Upgraded Knowledge Base to use local multilingual embeddings, enabling support for 50+ languages.
  • Auto-Translation Layer: Implemented real-time response translation, allowing the agent to reply in the user's native language.
  • Global Accessibility: Added native support for Chinese, French, Spanish, and Hindi interactions.
October 2025
Production Readiness & Advanced AI Integration
  • Bayesian Network Integration: Implemented a 7-node causal network for probabilistic SWE inference.
  • Comprehensive Data Strategy: Established a robust pipeline merging historical (2010-2020), simulated (2021-2024), and real-time (2025) data.
  • Enhanced Preprocessing: Deployed advanced imputation, outlier detection, and feature engineering with 15+ new metrics.
  • GitHub Standardization: Achieved full compliance with GitHub Spec Kit, including CI/CD pipelines and automated testing.
September 2025
Model Excellence & Real-time Compliance
  • 100% Real Data Compliance: Eliminated all hardcoded data, integrating live OpenWeatherMap API feeds from 8 monitoring stations.
  • GRU Ensemble Success: Achieved 88.52% R² score with an optimized GRU ensemble model, outperforming single models.
  • Production UI Redesign: Launched a professional, two-column dashboard with real-time charts and provenance tracking.
Initial Phase
Foundation & Core Development
  • System Architecture: Designed the scalable FastAPI backend and React/Vanilla JS frontend architecture.
  • Data Pipeline: Built the initial high-resolution data pipeline for Sentinel-2 and DEM data processing.
  • MVP Launch: Successfully deployed the initial version of the HydrAI-SWE system for the Red River Basin.