AI agents · IoT · Built
Digital Twin Grow Room Assistant
An agentic AI system for real-time environmental monitoring of a grow room, with LSTM-based yield forecasting and reinforcement-learning recommendations for actuators.
The problem
A grow room has many sensors and actuators, and someone has to watch them and decide what to change. This agent watches conditions in real time, forecasts yield and recommends actuator settings.
How it works
- 01IoT sensor readings
- 02LSTM yield forecast
- 03LangGraph agent
- 04RL actuator recommendations
- 05Chat interface
Highlights
- Agent built with LangGraph and Llama 3.3 70B served through Groq
- Real-time IoT monitoring, LSTM yield forecasting and RL-driven actuator recommendations
- OpenWeather and Tavily Search integrations, with multi-turn memory through the LangGraph MemorySaver checkpointer
- Gradio chat interface; built for an M.S. thesis use case
Built with
More like this
- CMDF-NetCross-modal deepfake detection
- Financial Analyst CrewMulti-agent research app on CrewAI
- Multi-Capability AI AssistantTool-calling agent in Python
Need something like this?
I build AI agents, automation and forecasting for small businesses and teams. Read about ai agents for small business, or see how I work with clients.