I will audit your rag retrieval quality and identify regressions

I
isaqueborges
I
isaqueborges
Isaque Borges
Sommige informatie wordt in het Engels weergegeven.

Over deze dienst

Is your RAG pipeline returning worse results after an embedding, reranking, or retrieval change?

I provide independent, reproducible RAG retrieval quality audits for AI teams and agencies.

For up to 30 queries and two comparable retrieval configurations, I analyze baseline vs. candidate rankings, identify query-level regressions and recoveries, and investigate missing or displaced evidence.

You receive a technical report, machine-readable findings, and reproducibility details.

My public audits include RouteMind and DPOLens. In RouteMind, I identified 15 scoring discrepancies across 1,400 query/arm results. The maintainer independently reproduced the findings and corrected the benchmark.

Required inputs include a fixed query set, relevance judgments or expected evidence, and compatible retrieval exports.

This service does not include implementing code fixes or evaluating generated answer quality.

Please contact me before ordering to confirm compatibility and scope.

Maak kennis met Isaque Borges

Isaque Borges

RAG Retrieval and Evaluation Specialist

  • Afkomstig uitBrazilië
  • Lid sindsokt 2026
  • Gem. reactietijd1 uur
  • Talen

    Engels
I specialize in independent RAG retrieval audits, benchmark validation, and regression analysis. I help AI teams identify retrieval failures, compare pipeline versions, and produce reproducible technical reports. My public work includes RouteMind and DPOLens audits. In RouteMind, I identified 15 scoring discrepancies independently reproduced by the project's maintainer, leading to benchmark corrections. I focus on clear evidence, transparent methodology, and actionable recommendations.