Dan S

* Design and deployment of an automated evaluation solution for a GenAI model, intended for product and data science teams * Definition of the target architecture (evaluation pipeline, orchestrator, results storage, monitoring dashboard) in collaboration with tech teams * Implementation of a fine-grained model performance analysis approach: extraction of key metrics, error categorization, relevance scoring * Introduction and implementation of a model-based evaluation method (LLM as a Judge), enabling large-scale benchmarking of multiple model versions * Close collaboration with product, data, and ML teams to integrate evaluation into model development and iteration cycles * Contribution to the definition of internal guidelines on GenAI model interpretability and reliability

Adoption and implementation of AI and generative AI models to develop innovative solutions tailored to the specific needs of our clients across various sectors.

Supervision of all processes related to data, from defining the roadmap to the development and industrialization of data and AI projects.

Responsible for detecting malicious activities by leveraging variance analysis models applied to globally collected network security data.

Implementation of analytical and data solutions for various clients such as Peugeot, Citeo, Savencia, and Europcar.

Improving the probability of authenticity in bank withdrawals by closely collaborating with business experts and developing tailored machine learning models.