The internal tool behind a 15,000-paper knowledge collection.

A retrieval system is only as useful as the documents behind it. DossieClean gives Dossie an internal workflow for turning academic and research PDFs into reviewed metadata, configurable text chunks, and searchable embeddings.

Why it exists

DossieClean did most of the document-preparation work behind DossieScholar’s collection of 15,000 research papers from Addis Ababa Science and Technology University. The work connected PDF preparation and metadata review to the collection’s searchable knowledge foundation.

15,000 research papers

Single and batch uploads extract titles, authors, publication years, abstracts, and keywords with GPT-4o-mini. The metadata preview lets an operator correct those fields before saving, keeping a human review step in the ingestion process.

Review before storage

PyPDF2 extracts PDF text. Configurable chunk size and overlap control how that text is divided; OpenAI text-embedding-3-small produces 1,536-dimension embeddings. Azure Blob Storage holds the PDF files while Qdrant holds the vector records.

A retrieval-ready pipeline

The dashboard shows document statistics and recent records. Natural-language search can target documents or chunks, while document details expose metadata and individual chunks. Batch processing includes real-time progress tracking, and settings expose chunking controls alongside dark and light themes.

Working with a collection

The broader DossieScholar workflow included Python and FastAPI pipelines to clean, process, index, and embed the research collection, OCR digitization, and vector retrieval with RAG. API and concurrent-user tests with Locust helped identify bottlenecks and improve reliability. DossieClean supplied the internal preparation workflow within that larger system.

The wider research workflow

The application uses Python 3.8+, Flask, Jinja2 templates, and JavaScript. Its API supports metadata extraction, reviewed saves, direct processing, batch execution, and batch-status checks. This project is presented as Dossie’s internal tooling rather than a separate public product.

Built as an internal utility

DossieClean

Dossie’s internal document-preparation tool, responsible for much of the processing behind DossieScholar’s collection of 15,000 research papers. It turns PDFs into reviewed metadata, text chunks, and embeddings for semantic retrieval.

Add one PDF or a batch of research documents.

Upload

Extract text and AI-generated metadata, then review or edit the fields.

Extract & review

Apply configurable chunking and generate 1,536-dimension vectors.

Chunk & embed

Save PDFs to Azure, index vectors in Qdrant, and search the collection.

Store & retrieve

Connecting the knowledge inside an organization.

Useful institutional knowledge is often fragmented across physical records and disconnected digital sources.

The opportunity

Designed an AI knowledge platform to bring that information together for search, retrieval, and practical use.

The system

Separately, at Dossie Technologies, validated demand for an AI medical-record digitization workflow through research across 147 hospitals and health centers with 167 healthcare professionals. These figures describe research coverage, not customers or deployments.

Grounded in research

Dossie

An enterprise AI knowledge platform designed to connect physical records, documents, and databases into a centralized, searchable knowledge base.

Connect the product concept to physical records, documents, and databases.

Bring sources together

Organize fragmented institutional information into a searchable foundation.

Create a knowledge base

Design for search, retrieval, and practical use of that knowledge.

Dossie.

The opportunity

The system

Grounded in research

How it comes together