Result listed — see original announcement
Original award wording
Finalists
Project profile
Zeppelin
Zeppelin is an S3-native vector search engine that separates storage from computation. The creator describes storing vectors in object storage and using stateless, coordination-free query nodes to scale horizontally, rather than maintaining replicated in-memory index clusters.
The creator reports long agentic coding sessions with Claude Opus 4.6 and TLA+ specifications for concurrency issues. The current README adds full-text search and reports local-loopback MinIO benchmarks, which are not cloud S3 measurements or verified competition-time features. It explicitly describes a pre-1.0 project; the submission's production-grade wording is not independent proof. The software was not run.
Creator or team (original public names)
Anup Ghatage
Links & sources
Links may have changed since we checked them. Descriptions from creators have not been independently tested.
Details in the original language
Original title
Zeppelin
Original creator credit
Anup Ghatage
作品是什么
S3 原生向量搜索引擎；当前 README 另列全文搜索等扩展，不能倒推全部属于赛时版本。
给谁用
需要部署向量搜索、管理对象存储检索服务的开发者。
问题或体验
当前 README 将问题表述为常规向量数据库需要维持复制与内存索引节点集群，成本不随查询量同步降低；本作尝试拆开存储和计算。
如何使用
作者说赛时通过 Claude Opus 4.6 长时间 agentic 编程会话进行规划、实现、调试，并用 TLA+ 规格查找并发问题；引擎把向量存于 S3，以无需协调的节点水平扩展。
得到什么
向量数据库软件与公开仓库。当前文档的设计特点是对象存储为事实来源、节点无状态；当前性能数字为本地回环 MinIO 测量，不是云端 S3 性能实测。
关键机制
对象存储保存向量、无状态节点查询，拆分存储与计算。当前文档的性能数据为本地 MinIO 环境，尚未复验。
特点
重点是对象存储为事实来源、计算节点无状态；赛时向量检索与当前 README 的全文检索扩展分开记录。
Events & award announcements
Read original announcement
Sources last checked: 2026-10-11