SkillGraph

GraphRAG skill graph for reusing diagnostic reasoning from agent traces. First author, MLSys 2027 submission.

SkillGraph: Reusable Reasoning Skills from LLM Agent Traces

Summer 2026 – Present · First author

Advisors: Dr. Tianyin Xu (University of Illinois Urbana-Champaign) and Dr. Ali Anwar (University of Minnesota)

Leading development of a GraphRAG-based skill graph for extracting and reusing diagnostic reasoning from agent trajectories.

  • Built a database of reasoning traces for skill extraction and retrieval on new system incidents.
  • Identified shortcut behavior that corrupts learned skills and designed a process-level critic that evaluates reasoning quality rather than final outcomes alone.

Impact: Submission at MLSys 2027.

(Khan & others, 2027)

References

2027

  1. SkillGraph: Reusable Reasoning Skills from LLM Agent Traces
    Saman Bahzad Khan and others
    In Conference on Machine Learning and Systems (MLSys), 2027
    Under Submission. First author.