diff --git a/.env.template b/.env.template
index 48fe88a..77bfb1f 100644
--- a/.env.template
+++ b/.env.template
@@ -1,26 +1,19 @@
-GOOGLE_API_KEY=
-
-# To use vertexai keep it true and false to use gemini
-GEMINI_USE_VERTEX=true 
+GOOGLE_API_KEY=AIzaSyBSW2NPqACOJnvm396j6nt3PydiEwJ2-HU
+GEMINI_USE_VERTEX=false
 
+# You can leave other variables blank for local dev if not using Vertex AI
 GCP_PROJECT_ID=
-GCP_REGION=      
-
-# BigQuery 
+GCP_REGION=
 BQ_DATASET_ID=
 BQ_TABLE_ID=
-BQ_LOCATION=                 
-
+BQ_LOCATION=
 INDEX_ENDPOINT_ID=
 DEPLOYED_INDEX_ID=
 
-
 EMBED_MODEL_NAME=nomic-ai/nomic-embed-text-v1.5
 INDEX_ENDPOINT_ID_FULL=
 VECTOR_API_EP=
 
-
-# For Scraping data
 PUBLIC_API_URL=https://api.knowledge-space.org
 ELASTIC_BASE_URL=
 ELASTIC_USERNAME=
diff --git a/README.md b/README.md
index 00f71e0..73c2915 100644
--- a/README.md
+++ b/README.md
@@ -145,6 +145,7 @@ The backend requires specific environment variables to connect to **Google Cloud
 In one terminal, from the project root with the virtual environment active:
 
 ```bash
+cd backend
 uv run main.py
 ```
 
diff --git a/backend/retrieval.py b/backend/retrieval.py
index 866d6c9..f89a0ce 100644
--- a/backend/retrieval.py
+++ b/backend/retrieval.py
@@ -233,3 +233,85 @@ def search(
         except Exception as e:
             logger.error(f"Matching Engine search failed: {e}")
             return []
+        
+    
+    def hybrid_search(
+        self,
+        query: str,
+        top_k: int = 20,
+        context: Optional[Dict[str, Any]] = None
+    ) -> List[RetrievedItem]:
+        """
+        Hybrid retrieval: Vector Search + Keyword Search with RRF
+        """
+
+        vector_results = self.search(query, top_k, context)
+        keyword_results = self.keyword_search(query, top_k)
+
+        if not keyword_results:
+            logger.info(
+                "Hybrid search fallback: returning vector results only"
+            )
+            return vector_results
+
+        fused_results = reciprocal_rank_fusion(
+            vector_results,
+            keyword_results
+        )
+
+        logger.info(
+            
+            "Hybrid search results | vector=%d keyword=%d fused=%d",
+            len(vector_results),
+            len(keyword_results),
+            len(fused_results),
+        )
+
+        return fused_results[:top_k]
+
+
+
+
+
+    def keyword_search(
+        self, query: str, top_k: int = 20
+    ) -> List[RetrievedItem]:
+        """
+        Keyword-based search (BM25 / Elastic).
+        Placeholder implementation for hybrid search.
+        """
+        logger.info("Keyword search called (placeholder)")
+        return []
+
+
+
+def reciprocal_rank_fusion(
+    vector_results: List[RetrievedItem],
+    keyword_results: List[RetrievedItem],
+    k: int = 60
+) -> List[RetrievedItem]:
+    """
+    Combine vector and keyword results using Reciprocal Rank Fusion (RRF)
+    """
+
+    scores: Dict[str, Dict[str, Any]] = {}
+
+    def add_scores(results: List[RetrievedItem]):
+        for rank, item in enumerate(results):
+            if item.id not in scores:
+                scores[item.id] = {
+                    "item": item,
+                    "score": 0.0
+                }
+            scores[item.id]["score"] += 1 / (k + rank + 1)
+
+    add_scores(vector_results)
+    add_scores(keyword_results)
+
+    fused = sorted(
+        scores.values(),
+        key=lambda x: x["score"],
+        reverse=True
+    )
+
+    return [x["item"] for x in fused]