diff --git a/src/main.py b/src/main.py
index 8de53f1..cf7ecb8 100644
--- a/src/main.py
+++ b/src/main.py
@@ -1,3 +1,15 @@
+
+from fastapi import FastAPI, HTTPException
+from fastapi.responses import JSONResponse
+from pydantic import BaseModel, Field
+from typing import List
+import numpy as np
+from sklearn.linear_model import LogisticRegression
+from sklearn.model_selection import train_test_split
+from sklearn.preprocessing import StandardScaler
+from sklearn.pipeline import Pipeline
+from typing import Dict, List
+
 import os
 from fastapi import FastAPI
 from fastapi.responses import JSONResponse
@@ -31,6 +43,7 @@
 )
 from src.fraud import check_fraud_rules
 
+
 app = FastAPI(
     title="Veritix Microservice",
     version="0.1.0",
@@ -40,6 +53,87 @@
 logger = logging.getLogger("veritix")
 
 # Global model pipeline; created at startup
+
+model_pipeline: Pipeline | None = None
+
+mock_user_events: Dict[str, list[str]] = {
+    "user1": ["concert_A", "concert_B"],
+    "user2": ["concert_B", "concert_C"],
+    "user3": ["concert_A", "concert_C", "concert_D"],
+    "user4": ["concert_D", "concert_E"],
+}
+
+class PredictRequest(BaseModel):
+    """Request body for /predict-scalper endpoint.
+
+    Each record represents aggregated event signals for a buyer/session.
+    """
+    features: List[float] = Field(
+        ..., description="Feature vector: e.g., [tickets_per_txn, txns_per_min, avg_price_ratio, account_age_days, zip_mismatch, device_changes]"
+    )
+
+
+class PredictResponse(BaseModel):
+    probability: float
+
+class RecommendRequest(BaseModel):  
+    user_id: str  
+
+
+class RecommendResponse(BaseModel):  
+    recommendations: List[str]
+
+def generate_synthetic_event_data(num_samples: int = 2000, random_seed: int = 42) -> tuple[np.ndarray, np.ndarray]:
+    """Generate synthetic data for scalper detection.
+
+    Features (example semantics):
+    0: tickets_per_txn (0-12)
+    1: txns_per_min (0-10)
+    2: avg_price_ratio (0.5-2.0)
+    3: account_age_days (0-3650)
+    4: zip_mismatch (0 or 1)
+    5: device_changes (0-6)
+    """
+    rng = np.random.default_rng(random_seed)
+
+    tickets_per_txn = rng.integers(1, 13, size=num_samples)
+    txns_per_min = rng.uniform(0, 10, size=num_samples)
+    avg_price_ratio = rng.uniform(0.5, 2.0, size=num_samples)
+    account_age_days = rng.integers(0, 3651, size=num_samples)
+    zip_mismatch = rng.integers(0, 2, size=num_samples)
+    device_changes = rng.integers(0, 7, size=num_samples)
+
+    X = np.column_stack([
+        tickets_per_txn,
+        txns_per_min,
+        avg_price_ratio,
+        account_age_days,
+        zip_mismatch,
+        device_changes,
+    ])
+
+    # True underlying weights to simulate scalper probability
+    weights = np.array([0.35, 0.45, 0.25, -0.002, 0.4, 0.3])
+    bias = -2.0
+    logits = X @ weights + bias
+    probs = 1 / (1 + np.exp(-logits))
+    y = rng.binomial(1, probs)
+    return X.astype(float), y.astype(int)
+
+
+def train_logistic_regression_pipeline() -> Pipeline:
+    """Train a simple standardize+logistic-regression pipeline on synthetic data."""
+    X, y = generate_synthetic_event_data()
+    X_train, _, y_train, _ = train_test_split(X, y, test_size=0.2, random_state=123)
+    pipeline = Pipeline(
+        steps=[
+            ("scaler", StandardScaler()),
+            ("clf", LogisticRegression(max_iter=1000, solver="lbfgs")),
+        ]
+    )
+    pipeline.fit(X_train, y_train)
+    return pipeline
+
 model_pipeline: Any | None = None
 etl_scheduler: "BackgroundScheduler | None" = None
 
@@ -88,6 +182,39 @@ def health_check():
 
 @app.post("/predict-scalper", response_model=PredictResponse)
 def predict_scalper(payload: PredictRequest):
+    if model_pipeline is None:
+        return JSONResponse(status_code=503, content={"detail": "Model not ready"})
+    features = np.array(payload.features, dtype=float).reshape(1, -1)
+    proba = float(model_pipeline.predict_proba(features)[0, 1])
+    return PredictResponse(probability=proba)
+
+# If you run this file directly (e.g., in a local development environment outside Docker):
+# if __name__ == "__main__":
+#     import uvicorn
+#     # Note: host="0.0.0.0" is crucial for Docker development
+#     uvicorn.run(app, host="0.0.0.0", port=8000)
+
+@app.post("/recommend-events", response_model=RecommendResponse)  
+def recommend_events(payload: RecommendRequest):  
+    user_id = payload.user_id  
+    if user_id not in mock_user_events:  
+        raise HTTPException(  
+            status_code=404,  
+            detail={"message": "User not found"}  
+        )  
+    user_events = set(mock_user_events[user_id])  
+    scores = {}  
+    for other_user, events in mock_user_events.items():  
+        if other_user == user_id:  
+            continue  
+        overlap = len(user_events.intersection(events))  
+        for e in events:  
+            if e not in user_events:  
+                scores[e] = scores.get(e, 0) + overlap  
+
+    recommended = sorted(scores, key=scores.get, reverse=True)[:3]  
+    return RecommendResponse(recommendations=recommended)  
+
     global model_pipeline
     try:
         if model_pipeline is None:
@@ -163,3 +290,4 @@ def on_shutdown() -> None:
             etl_scheduler.shutdown(wait=False)
         except Exception:
             pass
+
diff --git a/tests/test_recommend.py b/tests/test_recommend.py
new file mode 100644
index 0000000..3125967
--- /dev/null
+++ b/tests/test_recommend.py
@@ -0,0 +1,18 @@
+from fastapi.testclient import TestClient  
+from src.main import app  
+
+client = TestClient(app)  
+
+def test_recommend_events_valid_user():  
+    response = client.post("/recommend-events", json={"user_id": "user1"})  
+    assert response.status_code == 200  
+    data = response.json()  
+    assert "recommendations" in data  
+    assert isinstance(data["recommendations"], list)  
+    assert len(data["recommendations"]) <= 3  
+
+def test_recommend_events_user_not_found():  
+    response = client.post("/recommend-events", json={"user_id": "unknown"})  
+    assert response.status_code == 404  
+    data = response.json()  
+    assert data["detail"] == "User not found"  
\ No newline at end of file