All projects

Document collection and semantic comparison prototypes

Automating document collection, structuring content as JSON and experimenting with semantic comparison.

Professional experienceAutomation and data collection

An assignment from my professional background. No internal or confidential data is disclosed.

Abstract illustration of the main data flow — no internal or confidential data is shown.

Context

From April to June 2025, I worked within Apside’s Innovation team on several prototypes involving collection, automation and artificial intelligence.

  • Documents from public procurement platforms
  • CRM data
  • Unstructured content to transform into JSON

Challenge

The work addressed several needs: retrieving documents from public procurement platforms, structuring CRM data and experimenting with semantic content comparison.

Objectives

  • Automate public procurement document collection.
  • Structure CRM data to automate certain assignment tasks.
  • Transform unstructured content into JSON.
  • Experiment with semantic comparison using similarity measures and language models.

Contribution

I developed prototypes across these topics, working on collection, content transformation and comparison. The modular architecture was designed to make code maintenance and future changes easier.

  • Automating document retrieval with Python and Playwright.
  • Structuring CRM data and converting content into JSON.
  • Combining similarity measures and language models, including LLaMA, to compare content.
  • Organising the code in a modular architecture to support future changes.

Work carried out

Prototypes for automated collection, data structuring and semantic comparison. This exploratory work combined Python, Playwright, JSON and language models.

Technologies

  • Python
  • Playwright
  • LLaMA
  • JSON
  • GitLab

A similar challenge?

Facing a similar challenge? Let’s talk.

Describe your data, processes and objectives to discuss an engagement suited to your needs.

Discuss your project