Case Study
Battlefront Tactics
1st place Software Engineering Award · 2nd place AI tournament
Turn-based multiplayer strategy game, built over two semesters in a full Scrum process - 1st place at Ulm University's Software Engineering Award.
Context
Battlefront Tactics is a turn-based multiplayer strategy game, built as the official software engineering project at Ulm University - two semesters, one team, and a complete engineering process from requirements to release. The project competition is sponsored by Liebherr; our team won 1st place at the Software Engineering Award, plus 2nd place in the AI tournament, where the student teams’ autonomous agents played against each other.
The problem
A strategy game sounds like a gameplay problem, but the real challenge was an engineering one: six people, two semesters, a shared codebase - and a grade that depends as much on how you build as on what you build. Multiplayer synchronization, an AI mode with autonomous agents and a tournament system all had to fit into one coherent architecture that survived twenty-plus sprint iterations without collapsing.
How we built it
- Full Scrum, not Scrum theater. Sprints, plannings, reviews and retrospectives ran for the entire project. Requirements lived as user stories; every feature was estimated, prioritized and reviewed before merge.
- Godot + C# as the engine/language combination, with GitLab for merge requests, code review and CI.
- Three game modes on one core: classic multiplayer matches with unit placement, a tournament mode, and a separate AI mode in which autonomous agents play the game - the same rules engine drives all three.
- Technical documentation as a first-class deliverable: architecture decisions, interfaces and test strategy were documented and assessed.
Outcome
Working, complete software - menus, settings, unit management, multiplayer matches, AI tournaments - delivered on time and awarded first place across all participating teams. The AI agents took second place in the inter-team tournament.
What I took away
The biggest lesson wasn’t technical: a team that reviews honestly and retrospects without ego ships faster every sprint. On the technical side, building the AI mode against the same rules engine as the human mode forced clean separation between game state and presentation - a pattern I’ve reused in every project since, including digital-twin work.