Pierre Llompart

AI for drug discovery on AMD accelerators, in partnership with pharma and biotech R&D. - Research in agentic orchestration, co-scientist systems, large-scale co-folding, de novo design, and modelling. - Technical roadmap for AMD's drug discovery stack. - Scientific collaborations with pharma partners. - Benchmarking and performance optimization work on open-source life sciences models on ROCm.

Development and deployment of AI/ML solutions for computational drug design spanning small molecules and biologics: • Cross-Functional Collaboration: Close partnership with biologists, medicinal chemists, CMC teams, data scientists across departments and countries. Delivered training on AI tools, de novo design methods, and computational support to target project. • Protein-Ligand Binding Affinity: Developed in-house GNN with throughput >1M compounds/day; deployed for high-throughput virtual screening of docking, generative, and combinatorial libraries. • De Novo Protein Binder Design: Built end-to-end pipeline using denovo methods with comprehensive in silico screening for developability, binding affinity, and conformational stability assessment. • ADC Developability: Created novel GNN architecture for linker selection, DAR, aggregation, and stability prediction; model used to prioritize experimental constructs. • Antibody/Nanobody Developability: Developed ML models for viscosity, aggregation, HIC across mono/bispecific formats. Built internal descriptor module (300–6,000 features) for ML featurization and early alerts.

Industrial Ph.D. (CIFRE) combining research at the Laboratory of Chemoinformatics (Strasbourg) with applied drug discovery at Sanofi Integrated Drug Discovery. • Support industrial late-stage target projects by ML modeling. • Collaborate with on-site & international industrial wet-labs. • Selection & acquisition of commercial screening libraries for HTS. • Designed, led and managed research projects. • Mentored 4 Engineer interns on GenAI & Reinforcement Learning projects. • Curated by-hand hundreds of ADMET & Bioactivity datasets based on experimental conditions. • Developed novel Multi-Task Graph Neural Network covering +2,000 endpoints. • Virtual screening of large synthetisable libraries through combinatorial sampling and active learning • Presented my research at international conferences through posters and invited talks (American Chemical Society Fall 2024, Muséum d'Histoire Naturelle de Paris, Strasbourg Cheminformatics Summer School)

Pharmaceutical startup focused on natural compound-based therapies for neurological disorders. • Conducted ultra-large virtual screening of commercial libraries using ML, docking, and active learning for hit identification. • Developed Graph Neural Network for poly-pharmacological analysis integrating ChEMBL, PubChem, and STITCH protein network data. • Worked on poly-pharmaceutical projects exploring synergistic natural compound combinations for neurological disorder treatment.

• Designed scaffold-hopping protocols using probabilistic unsupervised learning for large-step molecular exploration. • Compared industrial in-house chemical space with ZINC and Enamine commercial libraries for structural enrichment. • Explored and compared industrial chemical spaces for structural enrichment. • Utilized unsupervised learning for target-based scaffold-hopping approach.

• Structure-based and ligand-based modeling of membrane ABC-transporter (ABCB10) for therapeutic hit discovery. • Homology modeling, ligand-protein docking, peptide-protein docking, full-atom MD simulations. • Optimized hit peptide sequence using Genetic Algorithm with ML-based binding affinity scoring.

• Studied monocyte differentiation signaling pathways from human blood samples. • Wet-lab experience: cell culture, assay preparation, data acquisition, temporal data analysis.

CIFRE Industrial PhD with Sanofi R&D Thesis: "Conception moléculaire par IA multitâche et exploration chémographique" (Molecular Design by Multi-Task AI and Chemographic Exploration) Supervisors: Prof. Alexandre Varnek, Dr. Gilles Marcou, Dr. Claire Minoletti Laboratory of Chemoinformatics, Faculty of Chemistry

Rank: 2nd/20 | High Honors (Mention Bien) Drug Design and Computational Biology: - Molecular Dynamics & Docking - Machine Learning & Cheminformatics - Organic & Medicinal Chemistry - Biostatistics & Bioinformatics

Bachelor of Science - Life Sciences & Chemistry September 2016 - July 2019 Rank: 3rd/125 | High Honors (Mention Bien) Foundational Knowledge in Biochemical Sciences: - Genetics & Molecular Biology - Enzymology & Biochemistry - Biophysics & Organic Chemistry - Metabolism & Animal Physiology