diff --git a/Optimization/Loss.py b/Optimization/Loss.py
new file mode 100644
index 0000000..e771d17
--- /dev/null
+++ b/Optimization/Loss.py
@@ -0,0 +1,39 @@
+import numpy as np
+
+class CrossEntropyLoss:
+    def __init__(self):
+        self.input = None
+        self.epsilon = np.finfo(float).eps
+    
+    def forward(self, input_tensor: np.ndarray, label_tensor: np.ndarray) -> float:
+        '''
+            forward pass of the cross entropy loss
+            Literature: H(P, Q) = -sum(P * log(Q)) : P is label_tensor, Q is input_tensor
+            Modification: 
+                added epsilon to avoid log(0) --> log(0) = -inf
+            Inputs: 
+                input_tensor: np.ndarray
+                label_tensor: np.ndarray
+            Expected Output: 
+                loss: float
+        '''
+        self.input = input_tensor
+        input_tensor = input_tensor[label_tensor==1]
+        loss = -np.sum(np.log(input_tensor + self.epsilon))
+        return loss
+        
+    
+    def backward(self, label_tensor) -> np.ndarray:
+        '''
+            backward pass of the cross entropy loss
+            Literature: dH/dQ = -P/Q : P is label_tensor, Q is input_tensor
+            Modification: 
+                added epsilon to avoid division by zero
+            Inputs: 
+                label_tensor: np.ndarray
+            Expected Output: 
+                gradient: np.ndarray
+        ''' 
+        gradient = -label_tensor/(self.input + self.epsilon)
+        return gradient
+        
\ No newline at end of file
diff --git a/Optimization/Optimizers.py b/Optimization/Optimizers.py
new file mode 100644
index 0000000..9901ed4
--- /dev/null
+++ b/Optimization/Optimizers.py
@@ -0,0 +1,19 @@
+import numpy as np
+
+class Sgd:
+    '''
+        Stochastic Gradient Descent optimizer class
+        
+        Equation:
+         w_new = w_old - learning_rate * gradient of loss function w.r.t weight tensor
+        
+        Position in the NN pipeline: 
+         After the backward pass to update the weights of the model
+    '''
+    def __init__(self, learning_rate: float) -> None:
+        self.learning_rate = learning_rate
+    
+    def calculate_update(self, weight_tensor: np.ndarray, gradient_tensor: np.ndarray) -> np.ndarray:
+        return weight_tensor - self.learning_rate * gradient_tensor
+    
+
diff --git a/Optimization/__init__.py b/Optimization/__init__.py
new file mode 100644
index 0000000..adf7b78
--- /dev/null
+++ b/Optimization/__init__.py
@@ -0,0 +1 @@
+__all__ = ["Optimizers", "Loss"]