Building a Golden Set for Production RAG
How to curate, refresh and use a high-signal RAG golden set: real queries, retrieval and answer labels, edge cases and drift refreshes.
GenAI Consulting25 min read
Insights
Building with Generative AI
Practical guides, architecture notes, and lessons from shipping GenAI to production.
Choosing Freshness SLAs for Enterprise RAG
How to set freshness targets for enterprise RAG and design sync, reindexing, cache invalidation and staleness budgets around each use case.
GenAI Consulting26 min read
Tenant-Aware Evaluation for Enterprise GenAI
How to build tenant-aware evals and release gates for multi-tenant GenAI, with leakage probes for RAG, caches, agents, memory and observability.
GenAI Consulting23 min read
Reranking in Production RAG Within Latency Budgets
How to add reranking to production RAG: architecture, model choices, candidate sizing, score fusion, evals and latency and cost tradeoffs.
GenAI Consulting22 min read
Canary Releases for Production GenAI Systems
How to canary prompt, retrieval, reranker, model and routing changes with traffic shaping, guardrails, evals and rollback triggers.
GenAI Consulting25 min read
Semantic Caches for RAG Without Data Leakage
How to design semantic caching for RAG: answer vs retrieval caching, ACL isolation, grounding revalidation, freshness invalidation and risk scoring.
GenAI Consulting25 min read
Prompt Optimization Without Fine-Tuning
How to treat prompt optimization as engineering: experiments, versioning, routing, evals, regression gates and observability before you fine-tune.
GenAI Consulting24 min read
Query Rewriting in Production RAG Systems
When to rewrite, expand or decompose queries in RAG, how to preserve intent and permissions, and how to evaluate the effect on retrieval.
GenAI Consulting23 min read
Session Memory Design for Production LLMs
How to design short-term and long-term memory for LLM apps: session state, summaries, profiles, external stores, privacy boundaries and debugging.
GenAI Consulting24 min read
Building an LLM-as-Judge Program for Evals
How to run LLM-as-judge evaluation that avoids bias, leakage and false confidence through calibration, cross-model checks and human review.
GenAI Consulting23 min read
Deterministic Preprocessing Pipelines for RAG
How to build RAG preprocessing that survives production: stable parsing, chunking, metadata, hashing, lineage, regression tests and rollout.
GenAI Consulting24 min read
Grounded Citation Pipelines for Enterprise RAG
How to build enterprise RAG citations that are verifiable and stable, with passage-level attribution, stable IDs, span alignment and validation.
GenAI Consulting25 min read
Showing 12 of 70 articles