Given an unweighted, directed graph with 10 million nodes and 100 million edges, explain how you would implement a solution to find the shortest path between two nodes that is faster than standard BFS for sparse graphs. Discuss heuristic approaches like Bi-directional BFS or A* and provide pseudocode for your chosen optimized method.