Context
Synchronous multiplayer word games demand instantaneous dictionary validation, synchronized room state across connected peers, and rapid Boggle matrix resolution without blocking server event loops.
What I built
- Designed an in-memory Prefix Trie structure in Python coupled with a Depth-First Search (DFS) traversal solver that indexes and evaluates 170,000+ words across a 4x4 matrix in under 10ms.
- Implemented real-time room orchestration and state broadcasting using Django Channels WebSockets.
- Built an interactive React and TypeScript frontend with live turn timers, submitted word tracking, and automated score tallies.
Technical approach
- The dictionary is loaded into memory as a trie node graph at server initialization, enabling O(L) prefix lookups (where L is word length) instead of repetitive O(N) list searches.
- The board solver executes a DFS with 8-directional neighbor exploration and cell-visited backtracking, pruning search paths immediately when a prefix is absent from the Trie.
- Django Channels handles WebSocket connection groups, broadcasting game transitions (lobby, round active, scoring, round end) to all players in a room without polling overhead.
- Client state manages local input buffers and word queues, reconciling confirmed submissions against WebSocket state broadcasts.
Security and verification
- Word validation occurs authoritatively on the backend Trie engine; client-submitted word lists are rejected if cells are non-contiguous or reused.
- WebSocket connection handlers validate room IDs and session payloads to prevent unauthorized state injection across active rooms.
- Unit tests verify Trie insertion, prefix matching, DFS board solving against known grid fixtures, and dictionary lookup edge cases.
Current limits
- In-memory Trie is held per-process in Python; multi-server scaling would require shared memory caching or distributed worker sync.
- Designed as an algorithmic practice and competitive word game MVP; persistent match leaderboards are pending future database backing.