Research overview#
My research focuses on AI-accelerated discovery and understanding of materials for electrochemical energy conversion. I combine:
- Physics-based modelling (DFT) to generate reliable energetics and structure insights
- Machine learning to generalize patterns and reduce computational cost
- Workflow automation to scale screening while keeping results reproducible
Current interests#
1) Electrocatalysis and interfacial chemistry#
- adsorption energetics and activity descriptors
- stability vs activity trade-offs
- realistic modelling choices (coverage, solvent approximations, uncertainty)
2) ML for atomistic systems (GNN potentials + property prediction)#
- structure-aware representations for surfaces and adsorbates
- in-domain vs out-of-domain detection
- calibrated ranking for high-throughput screening
3) Reproducible, scalable pipelines#
- automated dataset curation (deduplication, metadata, provenance)
- HPC-ready workflows (job arrays, checkpointing, failure recovery)
- transparent reporting: “what changed and why”
What I want to build next#
I’m excited by teams building AI4Science systems that connect:
- data generation (simulation/experiment)
- model training + evaluation
- decision-making (what to test next)
- interpretation (why it works)
If your group works on materials discovery, DFT/ML integration, or autonomous/closed-loop research, I’d love to connect.