--- title: Loop Engineering with Adaptive Parsing in Action:... description: An enterprise RAG pipeline uses an adaptive parsing cascade that starts with cheap PyMuPDF extraction and escalates to Azure Document Intelligence or a vision... canonical: https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM | daily.dev og:description: An enterprise RAG pipeline uses an adaptive parsing cascade that starts with cheap PyMuPDF extraction and escalates to Azure Document Intelligence or a vision... og:url: https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc og:image: https://api.daily.dev/og/posts/7noiwIbNC.png og:image:alt: Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM **[Towards Data Science](https://daily.dev/sources/tds)** · 25 min read · 0 upvotes · 0 comments ## Summary An enterprise RAG pipeline uses an adaptive parsing cascade that starts with cheap PyMuPDF extraction and escalates to Azure Document Intelligence or a vision LLM only when a page fails quality checks. Two real escalations are walked end to end: a flat-parsed table from the Attention paper routed to Azure DI, and an opaque architecture diagram routed to a vision model. An 18-run stress test reveals that LLM self-evaluation (context_structured flag) is unreliable as a standalone signal — wrong answers consistently assert clean structure — making deterministic upstream checks the primary load-bearers, with the LLM signal as a last-resort fallback. A post-generation groundedness check by a separate LLM catches fabrications the generating model misses. The data model stores parsing method per row, giving caching and auditability for free. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Loop Engineering with Adaptive PDF Parsing: Start Cheap, Pay for a Heavier Parser Only When the Page Needs It](https://daily.dev/posts/loop-engineering-with-adaptive-pdf-parsing-start-cheap-pay-for-a-heavier-parser-only-when-the-page-clzerc01t) · Towards Data Science · 0 upvotes · 0 comments - [Vision LLMs are PDF Parsers Too: Reading Charts and Diagrams for RAG](https://daily.dev/posts/vision-llms-are-pdf-parsers-too-reading-charts-and-diagrams-for-rag-4rms4zj8l) · Towards Data Science · 0 upvotes · 0 comments - [When PyMuPDF Can’t See the Table: Parse PDFs for RAG with Azure Layout](https://daily.dev/posts/when-pymupdf-can-t-see-the-table-parse-pdfs-for-rag-with-azure-layout-kgc3cizdx) · Towards Data Science · 0 upvotes · 0 comments - [Redesigning Banking PDF Table Extraction: A Layered Approach with Java](https://daily.dev/posts/redesigning-banking-pdf-table-extraction-a-layered-approach-with-java-r4y6b8lpf) · InfoQ · 0 upvotes · 0 comments --- Tags: [#rag](https://daily.dev/tags/rag) [View this post on daily.dev](https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM","url":"https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc"},"datePublished":"2026-07-20T15:32:15.057Z","dateModified":"2026-07-20T15:32:44.639Z","description":"An enterprise RAG pipeline uses an adaptive parsing cascade that starts with cheap PyMuPDF extraction and escalates to Azure Document Intelligence or a vision...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1f1b5a3e13779b44327d11f356a5faf3?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1f1b5a3e13779b44327d11f356a5faf3?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Towards Data Science","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Towards Data Science","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tds","url":"https://daily.dev/sources/tds"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/loop-engineering-with-adaptive-parsing-in-action-parsing-flat-tables-with-azure-and-figures-with-a--7noiwibnc","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"rag","timeRequired":"PT25M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Towards Data Science","item":"https://daily.dev/sources/tds"},{"@type":"ListItem","position":3,"name":"Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM"}]} ```