Engineer and business consultant in the energy sector, working at the intersection of industrial decarbonization and AI for materials discovery. My background is in electrochemistry (MSc thesis on electrodeposited catalysts for the hydrogen evolution reaction), and I now build ML and agentic tools that speed up R&D for low-carbon technologies, from catalyst screening to closed-loop optimization.
- ⚡ Specialized in electrochemistry, Bayesian optimization and AI-driven materials discovery
- 🤖 Building local-first agentic AI: RAG, evals, memory, MCP and prompt/context engineering
- 🎯 Fundamentally cross-disciplinary, mission-driven generalist who enjoys collaborative teams and creative solutions to hard problems
- 🌍 Here to contribute to the green transition
| Project | What it does |
|---|---|
| 💡 SALEM | SALEM gives literature-mining AI agents a way to check their own work: a Gaussian process predicts a trusted measured property (here, the PL peak of hybrid perovskites) from what the agent extracts, and the agent keeps rewriting its questions and database queries to lower that prediction error, in a self-reinforcing loop. |
| 🧪 catalyst-kg-agent | Knowledge-graph-grounded, cost-aware multi-agent system for green hydrogen catalyst screening on Materials Project data. Uses MACE & CGCNN surrogates, Bayesian optimization, pydantic-ai agents and MLflow tracking. |
| 🧠 notion-second-brain | Fully local RAG agent over my Notion workspace. Hybrid dense + sparse retrieval, cross-encoder reranking, Qdrant, quantized Ollama models and an eval harness with anchored rubrics. |
Languages
Data Science & ML
Materials Informatics
Bayesian Optimization
AI / LLMs & Agents - Context & harness engineering · RAG pipelines · Evals · Agent memory · MCP, Skills & Prompt optimization
Databases
Tools & IDEs
