Elias Limouni
Senior Data Scientist
Salary / Daily rate
Paris, France
Freelance
Remote only
Skills
Languages
About me
Intro
Senior Data Scientist with 4+ years of expertise in NLP, GenAI, and MLOps. Led AI transformation initiatives at Carrefour and eXalt, built multi-agent platforms serving 110+ users, and achieved 80% F1-scores on production NER models. Ready to deliver cutting-edge AI solutions that drive measurable business impact.
Work experience
PoC Factory Manager
eXalt Value
Sep 2024 - Present
2 years 2 months
Paris, France
Missions in the AI Lab of the eXalt group: - Spearheading the AI transformation journey by defining strategic roadmaps - Partnering with end users to identify high-impact Generative AI use cases, with a focus on agent-based systems - Guiding the development of cutting-edge PoCs and MVPs to bring ideas to life: an internal multi-agent platform to save efforts for the internal positions (sales, HR, C-level, talent acquisition...), with 110+ users - Empowering teams with tailored AI education — designing workshops, courses, and presentations for sales professionals, designers, managers, and executives - Building bridges across the organization through regular collaboration with key stakeholders to ensure alignment and success - Participating in expertise prospecting tasks alongside sales teams and executives, to bring new market opportunities
Senior Data Scientist
Carrefour
Sep 2024 - Present
2 years 2 months
91300 Massy, France
- Joining the Pricing team to work on the chaining algorithm of products - Conducting experiments, workshops and deployments to maintain and improve the current solution, while exploring new ones - Leading a GenAI experimentation with a Google Blackbelt to create a « Pricer » Gemini agent, then switching to a hybride architecture (classic ML algorithm + LLM for ambiguous cases) - GCP stack: Vertex AI (Datasets, AutoML…), Big Query (Data Form included), Composer
OPPSCIENCE
Jun 2020 - Oct 2024
4 years 5 months
Hybrid
Data Scientist & NLP Software Engineer
Aug 2023 - Oct 2024
1 year 3 months
Hybrid
- Relations Extraction with Few-Shot Learning Model, by exploring and optimizing state-of-art BERT vector combination techniques, reaching a minimum of 75% accuracy on complex real world data. - Making a NER Few-Shot model for a PoC, with deep debugging and analysis, reaching an overall ≈80% F1-score on OCRised production data (very noisy). - Deep Learning model in Java for Sentence Segmentation, to improve a custom rule-based NER solution. - Dataset Generation with auto-labelling using Generative AI (LLMs), to reduce the time-to-market by quickly fine tuning production models to customers-like documents - Designing and creating a whole pipeline about Active Data Generation (automating the Active Learning technique), to finetune a NER and a RE model automatically using a sequential augmentation technique with GenAI and a multi-agent approach until reaching threshold results
Data Scientist Junior
Sep 2020 - Aug 2023
3 years
Hybrid
- Multilingual NER models training and benchmarking, using TensorFlow and then PyTorch, reaching high scores (above 80% F1-score). - Various benchmark development to quantify several NLP models performances in Java and Python. - Pre/post processing of an OCR (based on Tesseract) optimization: rotation, scale, colors encoding to vector, improving the results by 10%. - Collection and benchmarking of Sentence Boundary Detection models, with several granularities, integration of the best one which improves the NER f1-score by 5% (higher impact on relation extraction). - Machine Translation (using models such as MarianMT) with MLOps tools to package this tool using Docker, and make it deployable and monitored quickly in production. - Using LLM for Data Generation and Pre-Annotation for custom and automated models training, which allows to quickly be able to generate documents to train a model for new clients, fastening the time to market. - Few-Shot learning models to quickly adapt solutions to the customers needs, again reducing the time to market. - NER models fine-tuning, quantifying how much documents needed to edit the tagset before going to production, allowing to pre-train models on a tagset and fine-tune it on a smaller set of data with a different tagset for each customer. - Documentation (Wiki) and oral presentations (creating PowerPoint presentations) in English, for my team and all the R&D teams, to present the work done, both on a technical and a business angle.
Data Scientist Intern
Jun 2020 - Aug 2020
3 months
Hybrid
- Classification modeling using Java and Python
Research Assistant
Netherlands eScience Center
Feb 2023 - May 2023
4 months
Amsterdam, Netherlands
Assistance of a researcher to study the capabilities of Distance Metric Learning for unsupervised Out-of-Distribution Detection (OOD). - Defining and developing the research questions, and the goals of the project. - Conducting a State-of-Art about the topic. - Implementing solutions and running experiments. - Writing reports in Latex for a latter integration in a publication, and Power-Points presentations. - Attending to research events, such as ICT.OPEN 2023 or the CWI Machine Learning Theory Boot Camp.
Co-Founder
Fidelight
Jun 2020 - May 2022
2 years
60200 Compiègne, France
In response to the COVID-19 crisis, my friend and I made the decision to establish Fidelight. Our primary objective was to offer local businesses a practical solution for enhancing customer engagement through a user-friendly digital loyalty system. - Meeting organization to list the needs of local business owners. - Drafting of specifications for the API. - Definition of tickets for issues and features, using a Gitlab Kanban board - Management of a freelance front-dev for the mobile app development. - Node.JS API development. - Database design for a future data-driven approach (UML drawing, complexity). - Google Cloud Platform implementations for the API hosting, using App Engine, Cloud Storage, SQL and VPC networks.
Communication Officer
Université de Technologie de Compiègne (UTC)
Sep 2018 - May 2022
3 years 9 months
60200 Compiègne, France
Student job done within the communication service of the UTC : - I made some videos to inform the students and make them aware of some main topics of the student life (for example the matter of the Human sciences for an engineer). - I made a lot of live videos during the events organized by the University or its associations.
IoT Developer Intern
FUSIOTECH
Jan 2019 - Mar 2019
3 months
Venette, Picardy, France · On-site
One-month internship part of the common core studies at the UTC. - Coding a program to connect automatically some sensors to a Raspberry Pi card, using the C++ language and some Linux scripts. - Miniaturization of an electronic board to monitor the air quality (using BME sensors). - Learning the welding using an oven - Debugging the card and the bug to improve the projects.
Education
ABC BootCamps
Entrepreneuriat / études entrepreneuriales
2024 - 2024
2 months
Université de Technologie de Compiègne (UTC)
Master of Science - MS
2020 - 2023
3 years
Université de Picardie Jules Verne (Amiens)
Bachelor of Technology - BTech
2019 - 2020
1 year
Université de Technologie de Compiègne (UTC)
Common Core
2017 - 2019
2 years 1 month