Ekimetrics
Documentation & Knowledge Manager — Audit and Agentic AI
Key Elements
| Category | Data |
|---|---|
| Organization | Ekimetrics |
| Period | March to July 2026 |
| Domain | Marketing Mix Modeling, Data & AI |
| Role | Documentation & Knowledge Manager |
| Assignment | Documentation audit, strategy, architecture, and Agentic AI |
| Audiences | Data Scientists, Data Engineers, consultants, and clients |
| Environment | Confluence, Sphinx, Python, reStructuredText, Rovo |
| Collaboration | Assignment through Crème de la Crème |
Project Overview
Ekimetrics develops advanced Data & AI solutions that help organizations understand performance drivers and optimize decision-making.
My assignment focused on the technical documentation supporting its Marketing Mix Modeling ecosystem, including the Data & AI Engine and the Modeling and Insight & Optimization applications.
The project combined documentation audit, target architecture, migration strategy, a Sphinx proof of concept, and the design of specialized Rovo agents for highly technical audiences such as Data Scientists, Data Engineers, consultants, and selected external users.
Context
The documentation was distributed across a knowledge base, PDF files, and a MkDocs website hosting Python notebooks. These resources contained advanced and valuable technical knowledge, but differed in structure, terminology, authoring methods, publication workflows, and levels of maturity.
The objective was to assess the documentation as a complete knowledge ecosystem and determine how it could evolve into a coherent website tailored to technical users working in a Python environment.
Key Deliverables
Documentation Strategy & Migration
I first conducted a comprehensive audit of the technical documentation.
The analysis covered information architecture, content structure, terminology, duplication, inconsistencies, functional coverage, readability, and contribution and publication workflows.
Based on the findings, I proposed a target architecture for a more unified, maintainable, and user-oriented documentation ecosystem. I also developed four short- and medium-term migration scenarios, ranging from incremental improvements to a broader documentation portal strategy.
To validate the recommendations, I created a Sphinx proof of concept in a Python environment. The POC demonstrated how a docs-as-code approach could support structured content, clearer navigation, reusable patterns, consistent rendering, and integration with development and deployment workflows.
Agentic AI for Documentation
In the second phase of the assignment, I designed four specialized Rovo agents and an orchestrator to improve access to, and the quality of, the documentation corpus.
The agents were designed to:
- search and synthesize technical information;
- support users in finding operational answers;
- identify redundant, duplicated, or inconsistent content;
- improve consistency in business and technical terminology;
- reduce ambiguity across documentation sets.
The orchestrator directed requests to the most relevant agent and combined their capabilities into a more coherent experience.
This work connected documentation quality, knowledge management, and AI readiness by turning the documentation into a more reliable source for both human users and AI systems.
My Role
As Documentation & Knowledge Manager, my role combined documentation strategy, technical analysis, information architecture, and AI experimentation.
I had to quickly understand a highly specialized Marketing Mix Modeling environment and work with documentation written for Data Scientists and Data Engineers.
This assignment illustrates my current positioning at the intersection of technical documentation, knowledge management, docs-as-code, AI-ready content, and Agentic AI.
Technologies & Tools
Confluence, Rovo, Sphinx, PyData theme, Python, reStructuredText, Visual Studio Code, GitHub, Sourcetree.
Confidentiality
The documentation, audit findings, recommendations, proof of concept, and Rovo agents are confidential and cannot be displayed publicly.
This portfolio page therefore describes the scope, approach, and expertise involved without disclosing proprietary content.
©Author: Florence Venisse, Technical Documentation & AI Expert – Updated version of 07/21/2026.