<p>Vehicle Routing Problem with Time Windows (VRPTW) is a common problem in logistics whose goal is to deliver goods to all customers while keeping the total travel distance as short as possible and minimizing the number of vehicles used. This paper proposes a new hybrid algorithm that is K-Means-NN-ALNS-SA, which combines four techniques. K-Means Clustering groups customers on the basis of their location so that the customers which are nearby are in the same cluster. This helps to make the problem shorter, simpler and faster to solve. Greedy Nearest Neighbour (NN) creates a delivery route by visiting the closest customer within time limit, while keeping in mind the capacity of the vehicle inside each cluster. Adaptive Large Neighbourhood Search (ALNS) provides improvement in routes by removing and rebuilding parts. Over time it understands which operator works best and selects the operator accordingly. Simulated Annealing (SA) prevents the algorithm from being stuck in the local best solution by sometimes accepting slightly worse solutions, in order to find the global best solution. This algorithm was tested using the Solomon benchmark dataset which consists of 56 instances – each instance having 100 customers and is commonly used for VRPTW research. The results show that K-Means-NN-ALNS-SA provides shorter distances of time travel and uses fewer vehicles. It performs better than traditional or single-method algorithms.</p>