Research Areas
I focus on AI systems that can be tested, adapted, and used in real workflows: benchmarks, knowledge layers, agent harnesses, post-training methods, and domain-facing applications.
Sustainability & ESG AI
Benchmarks, standards knowledge graphs, report-generation support, and climate-adaptation systems for auditable sustainability workflows.
ESG Benchmarks
Standards KGs
Report Generation
Climate Adaptation
Evaluation & Post-training
Acceptance tests, preference optimization, auditing, and productivity evidence for LLM systems that need defensible evaluation.
Acceptance Tests
Preference Optimization
Auditing
Post-training
Agent Infrastructure & Research Workflows
Harness layers, public agent ecosystems, research automation, and runtime reliability for agentic AI systems.
Harness Engineering
OpenClaw
Research Automation
Runtime Reliability
RecSys, Mobility, and Embodied AI
Sequential recommendation, activity recognition, embodied-agent evaluation, fall detection, and mobility-risk assessment.
Sequential RecSys
Mobility Risk
Activity Recognition
Embodied AI