GitMesh processes thousands of social signals every week—feature requests, bugs, competitor updates, and developer sentiment and turns them into ranked, actionable GitHub issues. Under the hood, OpenSearch is the backbone that makes this possible. This session walks through how we operate OpenSearch to support two demanding workloads: real-time indexing of unpredictable cross-platform signals, and fast search across developer profiles to match the right person to the right task.

You'll see how GitMesh manages versioned indices for zero-downtime migrations, dynamic field mapping for messy integration data, and nested documents representing both market signals and developer histories. We’ll break down our indexing pipeline (Postgres → SQS → workers → OpenSearch), our shard strategy, our caching and aliasing patterns, and the aggregation queries that connect market demand to engineering capacity. This is a practical, architecture-first look at operating OpenSearch in a system where search isn’t a feature, it is the product.