RAG and context engineering

Code: ENG-004 | Track: technical | Duration: 0.5 day

Building RAG pipelines, retrieval quality beyond naive vector search, and context engineering for production.

Teams whose retrieval quality is the bottleneck rather than the model.

Engineering and data teams

Python environment with API access; sample document set provided, or bring your own

Why context is the real bottleneck - context windows vs effective context, context rot, and where RAG sits among the alternatives.
RAG fundamentals and architecture - the pipeline from ingestion to generation, chunking strategies and handling messy real-world documents.
Build and break a baseline pipeline - hands-on construction of a working pipeline, then finding its failure cases and setting metrics.
Retrieval quality - beyond naive vector search - hybrid search, reranking, query transformation and permission-aware filtering, compared hands-on.
Context engineering techniques - assembling, compressing and caching context, and managing long conversations.
Evaluation, failure modes and production readiness - evaluating retrieval separately from generation, index freshness, access controls and budgeting.

Total time: 0.5 day (4 hours including two short breaks)
Participants: Up to 15
Delivery: Remote or in person (UK; other locations by arrangement)
What you leave with: A baseline-to-hybrid retrieval pipeline with eval metrics and a context-strategy decision framework