Freelancer profile translated to English.
Back to original languageAbout Miljan
Fractional Head of AI & Data | AI Engineer & Data Scientist
I transform the entire organization through AI (sales, marketing, finance, HR, operations, management), I interview teams, I map out how they truly work, and I eliminate manual processes one by one by building the internal tools that replace them.
The result in 12-18 months: a digital asset that belongs to you. An ecosystem of internal tools, connected to your real processes, where we gradually replace SaaS solutions costing €50-100k/year with internal solutions.
I decide the strategy and deliver the code myself. 8 years doing all the roles in a data department: first Data Scientist & Data Analyst, then Data Engineer to build my own data pipelines, and more recently AI Engineer to leverage the power of LLMs, with 30+ missions in production, from e-commerce with revenues up to several million euros to large groups (Mulliez, Sanofi, Puig).
Recent missions:
Celio: Lead Data Scientist, heading the Data Science & Analytics team. ML models in production for the entire company (AI product descriptions, email prediction, Marketing Mix Modeling, markdown optimization).
Wamiz: Outsourced Head of Data. Leading the data strategy and building the data platform from scratch in a few months (modern data stack: Airbyte, Dagster, dbt, BigQuery).
My method: I spend time with each business unit to identify pain points and the processes that cost the company the most. From this, I derive a roadmap prioritized by business impact, then we execute project by project, continuously, starting with what yields the fastest results. At each stage, change management to onboard teams into the transformation.
And I train your teams along the way: the goal is to make you self-sufficient, not dependent on me.
CIR approved by the Ministry of Research: for an R&D project, my services qualify for a 30% tax credit on the invoiced amount.
French
Native or bilingual
English
Fluent
Can work on-site
Paris (up to 50km)
Experience
- CELIOData Scientist Lead & AI Engineer — Retail Fashion BrandFASHION AND COSMETICSMarch 2026 - Today (5 months)Paris, FranceThe context: Celio, a retail fashion brand with a clear objective for my 9-month mission: implement data science and generative AI projects with a measurable ROI for each. No endless exploratory projects. Each initiative must prove its value.My scope is broad: lead on data science and AI engineering, but also on data analysis and complete customer analytics management. A role at the intersection of strategy and execution.The first projects involve V2s of existing high-potential tools:V2 of the AI product description application. The tool exists but needs to improve in quality and autonomy for the business teams. This is a subject I master perfectly, having built this type of application from scratch on a previous project.V2 of the email click prediction algorithm. The model predicts which customer will click on which campaigns. The goal: improve targeting and measure the impact on conversion rates.Behind these, structuring machine learning projects:Marketing Mix Modeling (MMM): modeling the impact of each marketing lever (media, CRM, promotions) on revenue to optimize budget allocation. A project with high strategic stakes that will directly influence management's budget decisions.Markdown optimization by ML: predicting and optimizing markdowns to preserve margins. In retail, markdown is one of the primary drivers of profitability loss.Each project follows the same logic: measurable objective, pre-defined success criteria, documented ROI.Stack: Python, Machine Learning, Generative AI, SQL
- Wamiz
On Malt
Head of Data (Externalized) — Wamiz (Pet media startup)PRESS AND MEDIAJanuary 2026 - Today (6 months)Paris, FranceThe context: Wamiz generates millions of page views per month but lacked a structured data strategy. Data existed scattered everywhere, but no one centralized or exploited it strategically. The CIO needed a Head of Data to lay the foundations, but not full-time. Hence the recruitment of an outsourced Head of Data via Malt.My role goes beyond technical aspects. I define the company's data strategy: governance, architecture, prioritization of data sources to centralize, and most importantly, how to make this data usable by the CEO and all business teams (editorial, marketing, product).What I implemented:Wamiz's complete modern data stack. Dagster for orchestration (each pipeline is monitored, each failure is traceable). BigQuery as the central data warehouse. dbt for data transformation and modeling (integrated tests, automatic documentation). Airbyte for ingesting all company data sources.The goal is not just to stack tools: it's to build a data platform that allows management to make decisions based on reliable data, and business teams to be self-sufficient in using that data.This is a Head of Data role in the strategic sense: defining what we collect, why, how we structure it, and who uses it.Stack: Dagster, BigQuery, dbt, Airbyte, GCP - EkosportAI Engineer, AI Agents — Ekosport (Outdoor retailer, 400+ suppliers)RETAIL (SMALL BUSINESS)November 2025 - February 2026 (3 months)Paris, FranceAI Engineer, AI Agents — Ekosport (Outdoor retailer, 400+ suppliers)The problem: The purchasing team spent hours manually re-entering catalogs from over 400 suppliers into the ERP. Each supplier sends their product matrices and purchase orders in different formats (PDF, Excel, free formats). 2.5 full-time equivalents were dedicated solely to this data entry.What I delivered:AI agents that automatically extract data from any supplier document (PDF, Excel, unstructured formats) and convert it to the ERP format. Extraction accuracy reaches nearly 100%, validated on an evaluation framework with a reference dataset and per-field metrics.A web application for the suppliers themselves: they upload their documents, and the AI pre-processes and detects missing or inconsistent information in real-time before the document reaches Ekosport. No more email back-and-forths.Integration with the product and inventory database to prevent stockouts upstream.Results: 2.5 FTEs freed up for value-added tasks, drastic reduction in supplier back-and-forths, improved inventory management.Stack: Python, LLMs (GPT-4, Claude, Gemini), LangChain, FastAPI, React, PostgreSQL, Document AIMalt Tags: Generative AI, LLM, Artificial Intelligence, Python, Langchain
Reviews
Tristan
CTO - Wamiz
Reviewed on 5/11/2026
Miljan has all the qualities we were looking for in a data professional: experience, proactivity, and autonomy.
He is assisting us with setting up our data stack following new company objectives (lead capture).
This includes: Dagster, GCP, BigQuery, dbt. Collaboration via Slack & GitHub.
He is a good communicator, he explains things clearly and adapts to his audience's level. He makes his work visible.
I highly recommend him, a very serious freelancer.
H
Hubert
Expert Bigquery, Dbt & CICD - SEIKI
Reviewed on 1/10/2026
Another excellent collaboration with Miljan for phase 3 of our project. The CI/CD implementation and the move to production went smoothly. This is the 3rd time I have used his services, and it is always a pleasure to work with someone so competent and reliable.
Recommendations
AD
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Master of Science in Managementemlyon business school2018Spécialisation en entrepreneuriat. Projet de fin d’études : l’impact de l’intelligence artificielle sur l’emploi
- Data ScientistOpenclassrooms + CentraleSupélec2019Compétences développées : -Analyser des données pour en extraire de la valeur et répondre à une problématique spécifique. -Modéliser le phénomène à l’origine de la création de données, à l’aide d’algorithmes de machine learning. -Mesurer et améliorer les performances de cette modélisation. -Automatiser des tâches telles que des recommandations, des prédictions, des classifications d’images, ... -Communiquer des résultats par la visualisation des données afin de faciliter la prise de décision. Expertises obtenues : -Traitement du langage naturel (NLP) : NLTK, TF-IDF, LDA, Word2Vec -Traitement d'images : OpenCV, Sift, Réseaux de neurones convolutifs (CNN) -Traitement de données séquentielles (finance, traduction) : LSTM, GRU, Seq2Seq, Attention, Transformer
Certifications
- Large Language Models (LLM) & TransformersSuper Data Science2024
- Describe and Clean Your DatasetOpenclassrooms2018