LLM Autoresearch Agent (NanoPath)

Paper: OpenReview

Benchmark: labless.dev · MedARC-AI/nanopath

• Drove 37% of the gain over baseline on NanoPath v1 (26% on v2) by building an LLM autoresearch agent with dual-branch search and anti-overfitting cycles (3rd-author paper, accepted at a NeurIPS 2026 workshop).
• Cut config errors 80% and token use 15× by designing agent infrastructure (knowledge graphs, SLURM, worktree isolation).

MedARC, 2026