NLP Research Engineer | Knowledge Graphs | LLMs & Ontology Engineering
I develop AI systems that bridge language models with structured knowledge — building Agentic LLM applications, benchmarking LLMs for various tasks
📍 Currently @ Ontology Engineering Group, Universidad Politécnica de Madrid.
-
Domain-Adaptive AI Systems: Developing LLM applications that serve as trustworthy collaborators in knowledge-intensive fields, from solar chemistry to ontology engineering with emphasis on explainability and user-centered design
-
LLM Evaluation & Benchmarking: Constructing comprehensive evaluation frameworks that measure not just accuracy, but explanation quality, scientific integrity, and real-world usability across diverse tasks and domains
-
Graph Learning Methodology: Exploring graph learning methods across diverse domains, with a particular focus on graph similarity search, node/link prediction, and adapting graph learning architectures for real-world problems such as software similarity.
Solar-QA — LLM-based RAG Pipeline for Solar Chemistry
Question-answering system helping solar chemistry experts review experiments from 700+ academic papers.
Stack: Python | RAG | Information Extraction | LLMs
SOEL LLMs for Ontology Engineering
Mapping LLM capabilities to ontology engineering tasks following LOT methodology.
Focus: Text2Triples, Triples2Onto, Ontology Generation
SoftSim — GNN Software Similarity Dataset
Novel dataset of 6,000+ software repositories for training graph neural networks on code understanding tasks.
Stack: PyTorch | GNN | Graph Learning
Languages: Python | SQL | CUDA
ML/DL: PyTorch | TensorFlow | Scikit-Learn | Transformers
Technologies: Docker | Unix/Shell | RAG Systems | Knowledge Graphs
Specialties: NLP | LLMs | Graph Learning | Ontology Engineering
- Email: ziyuan.wang@upm.es
- GitHub: @OEG-Clark
- Affiliation: OEG-UPM