Our Story
TerraLogic started off as a research project lead by Prof. Lixiang Jin with the department of Meteorological Sciences in CUHK. As an way of solution to the problem of offering precise and accurate local data which is critical for the Emergency response preparation in response to the numerous amounts of flooding incidents which has occurred in HK recently.

2020
Our first breakthrough came in 2020 with the development of our initial prediction algorithm. Working nights and weekends with limited resources, we managed to demonstrate a 12% improvement over traditional forecasting methods for rainfall predictions within a one-hour window. This early success allowed us to:
• Secure our first angel investment of $250,000
• Expand our team to include two additional data scientists
• Build relationships with local weather stations for data access
• Develop our first working prototype of the TerraForecast™ system

2022
Building on our early success, we focused on enhancing our AI models and proving their commercial viability. Despite operating on a shoestring budget, our team achieved remarkable progress:
• Improved our prediction accuracy to 82% for one-hour forecasts through advanced machine learning techniques
• Developed our proprietary data integration framework, combining satellite imagery, ground station readings, and radar data
• Successfully completed our airport pilot program, demonstrating a 40% reduction in weather-related delays

Current Situation
Today, TerraLogic stands at an inflection point. Our technology has proven its value and reliability through multiple real-world applications. We've demonstrated that our AI-driven approach to short-range weather prediction consistently outperforms traditional methods across various metrics:
• Enhanced our core AI model to handle microclimate variations, achieving hyperlocal predictions with 500-meter precision
• Developed an experimental mobile application with over 5,000 beta users providing real-world feedback
• Successfully predicted three severe weather events 2+ hours in advance during our pilot programs
