11/11/2026 updated
100 % available
AI specialist, Knowledge graphs, Semantic Web technologies, ML Infrastructure & Research Platforms
Grenoble, France
Only remote
PhD in Computer ScienceAbout me
Data Systems & AI Specialist with 8+ years of experience in knowledge graphs, semantic data modeling, ML, and scalable LLM infrastructure. Builds research-focused data platforms, GPU-accelerated pipelines, and intelligent tools, collaborating internationally with universities and UN agencies.
Fine-TuningData ArchitectureLinked DataMachine LearningNeo4jResource Description Framework (RDF)Semantic WebSPARQLHigh Performance ComputingChatbotsLarge Language ModelsMulti-Agent SystemsPrompt EngineeringApache SparkHuggingFace
Knowledge Graphs & Semantic Data Modeling
Expertise in knowledge graph architecture, RDF data modeling, SPARQL query generation and optimization, OBDA (Ontology-Based Data Access), ontology design and alignment, controlled vocabularies and taxonomies, as well as semantic web standards such as RDF, OWL, JSON-LD, and linked data. Production-scale knowledge graphs processing 12+ billion RDF triples have been designed and implemented.
ML Infrastructure & GPU Workload Deployment
Deep knowledge in deploying and optimizing large-scale LLMs across single and multi-GPU configurations (NVIDIA V100, A100, H100, H200, RTX family), leveraging tensor and pipeline parallelism. Inference serving with vLLM, Ray, Ollama, and llama.cpp, as well as model benchmarking, quantization, and distillation for models up to 650B+ parameters.
Data Pipeline Engineering
End-to-end pipeline design from raw data ingestion to production dashboards, including high-performance batch processing with Apache Spark on billion-scale datasets, ETL workflows for heterogeneous data sources, semantic annotation, and metadata enrichment. Architected pipelines for national-scale e-learning platforms processing billions of records.
CI/CD & Build Systems
Continuous integration and deployment pipelines, container image optimization (Singularity, Docker), build reproducibility and acceleration, Python packaging, and environment management across complex dependency trees.
Programming & Tooling
Primary language Python with deep ecosystem knowledge including packaging, virtual environments, testing, and profiling. Additional skills in SQL, shell scripting, and developer tooling and automation for research workflows.
Triplestores & Graph Databases
Hands-on experience with Virtuoso, AllegroGraph, GraphDB, Jena, Blazegraph (triplestores), Neo4j, and ArangoDB (graph databases), as well as GPU-accelerated RDF triplestore implementations using Apache Spark.
NoSQL Databases
Proficiency with Elasticsearch, OpenSearch, MongoDB, CouchDB, and ArangoDB for document storage, full-text search, and scalable data management.
AI/ML Domain Knowledge
Inference serving with vLLM, Ray, Ollama, and llama.cpp; model benchmarking and optimization including quantization and distillation; open-source LLM ecosystem covering Llama, Qwen, DeepSeek, and Mistral family; FastAPI-based OpenAI-compatible inference servers for NLP model deployment; agentic chatbots with autonomous SPARQL query generation.
Search Infrastructure
Design and maintenance of hybrid retrieval systems combining keyword and semantic search with reranking and automated entity extraction for large document collections.
Query languages
SQL, SPARQL, Prolog, Cypher
Experience
Professional experience5 - 10 years
Freelancer since2018
Languages
EnglishFluentFrenchBasic knowledgeSpanishNative speaker
Project history
Engineered a graph community network analysis pipeline for AGRIS using null models and stability checks. Containerized the Java-based AGROVOC tagging API for Kubernetes deployment. Owned deployment, monitoring, and maintenance of the GFAR web platform. Designed and built end-to-end data pipelines and interactive dashboards for agricultural indicators (Tapipedia). Maintained bibliographic repository with Dublin Core and Bibo ontology. Implemented a semantic annotation tool using AGROVOC covering 42,000+ concepts across 42+ languages. Collaborated fully remotely with distributed stakeholders across multiple time zones.
Designed and implemented an ontology for the LiSA learning progress tracking system for second-cycle medical education. Developed a comprehensive OBDA framework to translate relational data to RDF format. Built an autonomous agent for graph traversing 5 medical ontologies ( (UMLS,
SNOMED-CT-FR, MeSH, CIM-10 y LiSA). Deployed and optimized large-scale LLMs (up to 650B+ parameters) on OAR-managed GPU clusters. Architected end-to-end data pipelines for large-scale educational datasets, designed a hybrid retrieval system combining keyword and semantic search. Implemented an extension for the SPARQL query language to support embeddings search, outperforming the GraphRAG approach. Engineered an agentic pipeline (based on openweight models) for the generation of MCQ questions to improve medical reasoning.
Engineered multi-dimensional data models and ETL pipelines for international aid transparency (IATI) using RDF data vocabulary. Maintained the full geospatial platform stack including PostGIS and GeoServer, and developed an OBDA system using GeoSPARQL. Built and maintained search systems combining keyword and semantic retrieval with automated entity extraction. Designed AI-powered developer tools for non-technical stakeholders. Kubernetes infrastructure server management .Worked fully remotely with a globally distributed team.