RAG Pipeline Engineer - Design and evaluate production RAG retrieval pipelines
Designs, implements, and evaluates RAG pipelines for chunking, embeddings, hybrid retrieval, re-ranking, context assembly, and agentic retrieval.
Tags
Updated: 2026-10-04Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Design document chunking strategies
- Select corpus-validated embedding models
- Configure vector indexes
- Build hybrid search pipelines
- Implement metadata filtering
- Assemble retrieved context
- Integrate retrieval re-ranking
- Build retrieval evaluation harnesses
- Run retrieval ablation studies
- Monitor production retrieval quality
- Design agentic retrieval flows
- Implement query decomposition
- Add human-in-the-loop checkpoints
Inputs
- Source documents
- Document types
- Corpus samples
- Metadata filters
- Embedding model configuration
- Retrieval queries
- Golden evaluation datasets
- Latency and recall targets
Outputs
- Chunked documents
- Vector embeddings
- Configured vector indexes
- Hybrid search results
- Re-ranked contexts
- Evaluation metrics
- Retrieval quality logs
- Drift detection results
Requirements
- Python async runtime
- PostgreSQL with pgvector
- Embedding model access
- Vector database permissions
- Optional LangGraph support
- Optional evaluation framework support