Yash Agrawal

Digital twin control

Aqua Twin

A LangGraph water-management assistant that classifies natural-language requests, extracts simulation parameters, runs the water simulation, and analyzes results against stable ranges.

Open GitHub Back to systems
Aqua Twin artifact board with Streamlit chat, parameters JSON, LangGraph route, and stable-range gauges.
Artifact-style board from repository surfaces, not a product screenshot.
Status
Live repository
Proof type
Digital twin control
Stack
simulation Streamlit LangGraph Gemini digital twin
Source
github.com/rogue-socket/AquaTwin-Web

What it is

A conversational simulation graph

Aqua Twin routes chat requests through a LangGraph assistant. It can answer general water questions, analyze past simulations, or extract parameters and run a new simulation.

Why it exists

Simulation needs legible routing

The project makes the model's control role explicit: classify the user's intent, decide whether to simulate, pull out the exact water parameters, and explain the result against known stable ranges.

How it works

Classify, simulate, analyze

01Prompt
02Classify
03Simulate
04Analyze
  1. Classify requests as general query, past simulation, or run simulation.
  2. Extract water speed, temperature, dissolved oxygen, pH, salinity, and turbidity.
  3. Write simulation parameters and run the simulation path.
  4. Analyze output against stable ranges and return the result through Streamlit chat.