I'm an applied scientist working at the intersection of operations research and applied AI, focused on building decision-making systems that operate under real-world uncertainty.

My work blends optimization, simulation, and learning to produce systems that are robust, interpretable, and deployable—not just benchmark-friendly.

Some of my old projects can be found at https://github.com/xiaoxuanh

• Optimization & OR x AI

  • Stochastic optimization, planning, and simulation
  • Decision-making under uncertainty
  • Hybrid OR + ML systems that combine optimization, learning, and heuristics
    • Storage-with-DP: stochastic DP with embedded time series price model and clustering based state space management to achieve tractable optimal real-time energy storage control.

• Multi-Agent Systems

  • Designing end-to-end Multi-Agent environments and simulators
    • InvestESG: an environment studying how multiple AI agents behave and interact in a realistic and complex intertemporal social dilemma.
  • Modeling coordination, competition, emergent behavior, and ultimately AI safety
  • Studying system-level effects from agent interactions in shared environments

• Applied AI

  • Graph-based knowledge retrieval, particularly for code and structured reasoning
  • Deploying AI-based workforce scheduling systems and evaluating them against human-managed baselines
  • Building applied AI systems for personal productivity
    • jobflow-ai: automated job search, resume tailoring, and LinkedIn outreach planning that can be executed on a daily basis.

• Practical impact over toy results
• Robustness under noise, scale, and misspecification
• Using AI to augment human productivity and improve decision-making