Mounir M.

Data Scientist ML & MLOps Engineer, biochemist by training.

I design, train, and deploy Machine Learning models, real-time ML APIs, and LLM-based RAG systems into production. Certified Expert in Data Engineering and Data Science (Bac+5, RNCP39775 Level 7, Data Science option).

My added value comes from a rare dual expertise: nearly 9 years in scientific R&D and industrial quality control, including 5 years at Merck Serono (Geneva) on PK/PD pharmacokinetic studies and in silico predictive modeling (BBB crossing, 10 years of in vivo data), 1.5 years at Nestlé Research Center (Lausanne) on ISO 17025 LC-MS/MS method validation, and 1 year at Refresco in physicochemical analyses. This data practice in a demanding scientific environment nourishes my Data Scientist rigor today: feature engineering, drift control, applicability domain, regulatory traceability (GDPR, AI Act, FDA-EMA).

Specifically, I deliver:

· end-to-end ML pipelines (FastAPI, Docker, MLflow, DVC, GitHub Actions CI/CD),

· production-ready RAG systems (LangChain, FAISS, Mistral, RAGAS evaluation),

· AI project scoping (System Design, ROI, GDPR, AI Act, COMEX presentation).

My personal project DeFrag (defrag.eqqinox.dev), an LC-MS/MS spectral annotation pipeline filed with IDDN and deployed as a live API, demonstrates this ability to carry an ML project from raw data to autonomous production. Validated on 190k external GNPS spectra with an F1 drift of 0.004 between internal and external validation.

Full portfolio with all my projects: eqqinox.dev

Source code: github.com/Eqqinox

Bilingual French/English, based in Lyon, mobile and open to remote work.

Available for short or long-term assignments, on a daily rate or fixed-price basis.