I recently applied for a job that asked whether I'd be willing to complete a coding challenge. The challenge in question was listed as "hard" on the programming site and after what seems like 4 years only 26% of the developers who attempted it have been able to solve it.
I scored 100% - but I don't say that to brag. After completing the test I was able to view the solutions my peers created... and mine is certainly not as elegant as some of the ones I saw! That said, I'm still proud of my approach to a solution, and I thought it would be fun to share it here.
The core problem involves a turn-based simulation with six possible moves per turn and a potential of 30-50 turns. Thus, computing and storing all 1.41e+38 possible moves required to find the best move isn't feasible using a brute force search.
Therefore, the challenge requires the application of computer science concepts such as:
- recursion
- data structures (particularily, trees)
- search algorithms such as depth first search and backtracking.
- heuristics such as alpha-beta pruning
Quick summary
- You have a four lane bridge with one to four bikes at the start of the bridge.
- Your goal is to get a minimum number of bikes across the bridge.
- To complicate matters the bridge has pot holes that bikes have to navigate by moving left or right or jumping over them.
- Your challenge is to write a game AI that can get at least the minimum number of bikes accross the bridge.
- The programming site provides you a turn-based simulation that will allow you to test your AI.
See: Deeper summary of the core challenges
Given more free time I'd like to implement a version that uses a distributed search across multiple CPU cores. Although, I do think that would be overkill for this problem. It's just a bit painful to realize that only one of my machine's 10 cores was working on the problem.
| File | Description |
|---|---|
| source/bike_ai.py | The AI that computes the best next move for each turn. |
| source/simulator.py | A module that manages the states of the simulator and also processes and scores potential moves. Used by the bike_ai.py module and the simulation_tester.py module. |
| source/simulation_tester.py | A simulation tester that can be used to test the AI locally |
| source/simulations.py | A collection of test cases that can be used by the simulation_tester.py module. |
| source/state_search_space_tree.py | A modle that defines the state nodes and search space tree. |
| source/main.py | A script that is used to host the AI within a simulation. |
| source/graphviz.py | A module that can be used to generate a graphviz graph of the search tree. |
- Files in the
scriptsfolder are used to manage a docker stack that hosts the AI and the simulation. See the Docker instructions for more information. - Files in the
python-devfolder are used to build the docker image that is used by the docker stack.