AI Agent - Yield Allocation Service

FastAPI microservice providing AI-driven yield predictions and risk analysis for DeFi protocols.

Setup

python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

Train Models

Before running the API, train the prediction models:

python src/train_model.py

This will generate:

• models/yield_predictor.pkl - Model for predicting future yields
• models/risk_analyzer.pkl - Model for risk assessment

Run Server

uvicorn src.main:app --reload

API will be available at http://localhost:8000

API Endpoints

GET /recommendation

Returns AI-recommended allocation weights, yield predictions, and risk scores.

Response:

{
  "allocation": {
    "aave": 4000,
    "morpho": 3000,
    "spark": 2000,
    "uniswap": 1000
  },
  "predicted_yields": {
    "Aave": 0.072,
    "Morpho": 0.068,
    "Spark": 0.070,
    "Uniswap": 0.065
  },
  "risk_scores": {
    "Aave": 0.30,
    "Morpho": 0.25,
    "Spark": 0.35,
    "Uniswap": 0.40
  },
  "confidence": 0.85
}

GET /risk

Returns risk analysis for all protocols.

GET /yields

Returns yield predictions only.

Integration

The frontend and StrategyManager contract can call these endpoints to:

1. Display current predictions
2. Get allocation recommendations for on-chain rebalancing
3. Monitor risk levels

See docs/ARCHITECTURE.md for integration flow.