Antoine Lucas
Scientific Software & AI Engineer · R Shiny, LLM agents, MLOps
Pharma · Cosmetics · Biotech · Paris, France
Research engineer with 5+ years turning R&D problems into production software in pharma, cosmetics, and biotech. 20+ R Shiny applications in daily use by researchers, statistical modelling (design of experiments, mixed models, multi-omics), reproducible delivery (tests, CI/CD, containers) qualifiable in GxP environments, and LLM agents / MCP tools built on top of it.
Experience
Sep 2025 – present
- Own a catalog of 20+ R Shiny applications in daily use by researchers: clinical data visualisation, formulation automation, and data lifecycle workflows.
- Designed and deployed object-segmentation computer vision models for cosmetics formulation and regulatory R&D.
- Run the delivery platform (Git, Azure ML, Posit Connect, Databricks) with reproducible environments, versioned code, and documented handover.
May – Sep 2025
- Led internal upskilling training on Bayesian statistics with R for consultants.
- Built reusable internal R packages and Quarto reporting templates used across missions.
- Fine-tuned Phi-3-mini (3.8B) with LoRA (PEFT) for claim verification, from data preparation to evaluation.
Dec 2024 – May 2025
- Delivered end-to-end R and Python pipelines for tabular and large-scale omics data: genomics, transcriptomics, and metabolomics.
- Embedded as the sole data scientist in an 8-person startup team shipping a SaaS product (FastAPI backend, Vue.js front-end).
- Ran scientific discussions with agri-food partners on experimental design, analysis, and ML for multi-omics integration.
Jul 2023 – Nov 2024
- Sole R Shiny developer and Scrum Master: built R and Python applications that simulate manufacturing plant resource capacity across global sites, informing drug production planning.
- Owned delivery end-to-end — requirements, methodology, validation, workshops, UAT, handover — with backlog and sprint ceremonies across researchers, managers, and project leads.
Nov 2021 – Jul 2023
- Wrote analysis methodologies and statistical analysis plans for proof-of-concept clinical studies, programmed the analyses in R, and delivered reports for more than 100 studies.
- Characterised clinical phenotypes with multi-block, multivariate, and variable-selection approaches on exposome, biophysical, omics, and microbiome data.
- Proposed suitable statistical methods and built Bayesian-network and other graphical visualisations for clinical datasets, alongside a senior statistician.
Earlier experience
Feb 2021 – Aug 2021
Personalised hair-recommendation models built from customer data.
Feb 2020 – Aug 2020
Data management and non-parametric analysis of gut-microbiome clinical studies.
Sep 2019 – Jan 2020
Wearables and digital-health studies in R, from questionnaire design to results.
Education
2021
2021
Technical Skills
Certifications
GitHub Foundations — GitHub (2024)
Scrum Master — Agilbee (2023)
Languages
French (native)
English (fluent, professional)