Viewing offline recorded demonstration
Search is distributed.
See every step.
A custom distributed BM25 search engine built over a 250,000-document Wikipedia corpus. Explore the internal execution trace, topology, and performance metrics.
The Corpus
Inverted index mapping over a 250,000-document Wikipedia dataset using memory-mapped zero-copy access.
Sharding & Replication
8 logical shards distributed across 4 workers with a replication factor of 2 for failover and high availability.
Scatter-Gather
Concurrent dispatch to primary and replica shards via an intelligent coordinator that merges and sorts results globally.
Observability
Real-time Server-Sent Events (SSE) tracing every sub-millisecond execution phase from tokenization to merging.