Academy: open-source middleware for federated autonomous agents across HPC research infrastructure
BenBlaiszik · x · 2026-09-25
Ben Blaiszik detailed Academy, the open-source middleware behind his science-agent demo: built by Globus Labs researchers, it deploys stateful autonomous agents across distributed systems and federated research infrastructure.
- An agent is a Python object holding state, exposing remotely callable actions, and running its own control loops; an LLM is optional, not required.
- Existing agentic frameworks target cloud-native chatbots, while scientific work spans HPC systems, instruments, and institutions with separate queues, network policies, and identity providers—Academy is the execution and coordination layer for that setting.
- MIT licensed, pip install academy-py, v0.5.0; related work published at IPDPS 2026.
Related event: One Prompt Orchestrates Six MLIP Agents to Screen Battery Cathode Materials(2 posts)→
More from coding & agent
- LangSmith adds Trajectories, making long agent runs easier to analyze — LangChain · 2026-09-25
- Broken windows theory is 100% real in vibe-coded codebases — generativist · 2026-09-25
- Non-engineer shares multi-model agent setup: isolated Docker supervisor gates scope drift — alexcovo_eth · 2026-09-25
- Primus Society runs a society of autonomous AI researchers, claims 30% training boost — JayAlammar · 2026-09-25
- Gauntlet AI Opens Free 10-Week, Fully Funded AI Engineering Fellowship in Austin — petrusenko_max · 2026-09-25
- GitHub Security Lab open-sources an AI fuzzing pipeline that writes and tunes its own harnesses — GitHub Blog AI/ML · 2026-09-25