Codebase Semantic Indexing Service

You are a developer tools engineer. Build a codebase indexing service that gives AI coding agents precise semantic and structural retrieval. Tech Stack: [INSERT - Rust/Go indexer + gRPC API + Python/TS clients] Parsers: [INSERT - tree-sitter, LSP, SCIP/LSIF, ctags] Embeddings: [INSERT - OpenAI / Voyage / BGE code embedding models] Store: [INSERT - PostgreSQL + Qdrant/Milvus + object storage] Deliver the following modules in order: 1. Architecture: ingestion workers clone repositories, parse them into a symbol graph and chunks, emit embeddings; a serving API answers agent queries. 2. Data Model: repository snapshots, files, symbols (definitions/references/call graph), chunks with symbol-path metadata, an index version per commit. 3. Incremental Indexing: webhook/poll on Git push, reparse only changed files, patch the graph without full rebuild, tombstones and garbage collection. 4. Query APIs: semantic code search, symbol definition/reference lookup, callers/callees, find-similar-implementation, path and language filters. 5. Context Assembly: pack retrieved symbols with imports and signatures under a token budget; generate repository-level summaries and convention notes. 6. Testing: cross-language golden repos (Python/TS/Java/Go/Rust), recall/precision benchmarks, incremental consistency tests, performance SLO tests.