diff --git a/main.py b/main.py
new file mode 100644
index 0000000..1d035b2
--- /dev/null
+++ b/main.py
@@ -0,0 +1,162 @@
+from fastapi import FastAPI, HTTPException, Query
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel, Field
+from typing import List, Optional
+from langchain_openai import OpenAIEmbeddings
+from langchain_postgres import PGVector
+from dotenv import load_dotenv
+import uvicorn
+
+load_dotenv()
+
+# Initialize FastAPI app
+app = FastAPI(
+    title="Islamic Duas Semantic Search API",
+    description="Semantic search API for Islamic duas based on tags",
+    version="1.0.0"
+)
+
+# Add CORS middleware
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # Change to specific origins in production
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Database configuration
+CONNECTION_STRING = 'postgresql+psycopg2://postgres:test@localhost:5433/vector_db'
+COLLECTION_NAME = 'duas_tags_vectors'
+
+# Initialize embeddings and vector store
+embeddings = OpenAIEmbeddings(model="text-embedding-3-small")
+db = PGVector(
+    collection_name=COLLECTION_NAME,
+    connection=CONNECTION_STRING,
+    embeddings=embeddings
+)
+
+# Pydantic models
+class DuaResult(BaseModel):
+    id: Optional[str] = None
+    arabic: Optional[str] = None
+    transliteration: Optional[str] = None
+    translation: Optional[str] = None
+    urdu: Optional[str] = None
+    romanUrdu: Optional[str] = None
+    category: Optional[str] = None
+    occasion: Optional[str] = None
+    source: Optional[str] = None
+    tags: Optional[List[str]] = None
+    similarity_score: float
+
+class SearchResponse(BaseModel):
+    query: str
+    results_count: int
+    results: List[DuaResult]
+
+class SearchRequest(BaseModel):
+    query: str = Field(..., description="Search query", example="protection from evil")
+    k: int = Field(5, description="Number of results to return", ge=1, le=50)
+
+# Health check endpoint
+@app.get("/", tags=["Health"])
+async def root():
+    return {
+        "message": "Islamic Duas Semantic Search API",
+        "status": "running",
+        "endpoints": {
+            "search_get": "/search?query=protection&k=5",
+            "search_post": "/search (POST)",
+            "health": "/health"
+        }
+    }
+
+@app.get("/health", tags=["Health"])
+async def health_check():
+    try:
+        # Test database connection
+        test_results = db.similarity_search_with_score("test", k=1)
+        return {
+            "status": "healthy",
+            "database": "connected",
+            "embeddings": "loaded"
+        }
+    except Exception as e:
+        return {
+            "status": "unhealthy",
+            "error": str(e)
+        }
+
+# GET endpoint for search
+@app.get("/search", response_model=SearchResponse, tags=["Search"])
+async def search_duas_get(
+    query: str = Query(..., description="Search query", example="protection from evil"),
+    k: int = Query(5, description="Number of results", ge=1, le=50)
+):
+    """
+    Search for duas using semantic similarity on tags.
+    
+    - **query**: Your search query (e.g., "morning prayers", "protection from evil")
+    - **k**: Number of results to return (1-50)
+    """
+    try:
+        results = db.similarity_search_with_score(query, k=k)
+        
+        duas_results = []
+        for doc, score in results:
+            result = DuaResult(
+                id=doc.metadata.get('id'),
+                arabic=doc.metadata.get('arabic'),
+                transliteration=doc.metadata.get('transliteration'),
+                translation=doc.metadata.get('translation'),
+                urdu=doc.metadata.get('urdu'),
+                romanUrdu=doc.metadata.get('romanUrdu'),
+                category=doc.metadata.get('category'),
+                occasion=doc.metadata.get('occasion'),
+                source=doc.metadata.get('source'),
+                tags=doc.metadata.get('tags'),
+                similarity_score=round(1 - score, 4)
+            )
+            duas_results.append(result)
+        
+        return SearchResponse(
+            query=query,
+            results_count=len(duas_results),
+            results=duas_results
+        )
+    
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Search failed: {str(e)}")
+
+# # Get all unique categories
+@app.get("/categories", tags=["Metadata"])
+async def get_categories():
+    """
+    Get all unique categories from the duas collection.
+    """
+    try:
+        # This is a simple implementation
+        # For better performance, you might want to cache this
+        results = db.similarity_search_with_score("", k=1000)
+        categories = set()
+        for doc, _ in results:
+            category = doc.metadata.get('category')
+            if category:
+                categories.add(category)
+        
+        return {
+            "categories": sorted(list(categories)),
+            "count": len(categories)
+        }
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Failed to fetch categories: {str(e)}")
+
+if __name__ == "__main__":
+    uvicorn.run(
+        "main:app",  # Change "main" to your filename if different
+        host="0.0.0.0",
+        port=8899,
+        reload=True
+    )
\ No newline at end of file