Join Us
Projects

Real-world testbeds for agent-native systems.

Real-world experimental systems for validating agent-native infrastructure, orchestration, and trust.

01Exploratory

Agent-Native HPC

Autonomous AI agents for scientific workflows, HPC resource orchestration, and intelligent computing environments.

Agent-Native HPC explores how autonomous agents can assist, orchestrate, and optimize scientific workflows and high-performance computing environments.

Research Themes

Scientific workflow orchestrationHPC scheduler interactionResource-aware agent executionExperiment automationMonitoring and observability

Expected Artifacts

  • Agent-native HPC prototype
  • Scheduler-aware agent interface
  • Scientific workflow orchestration demo
02Exploratory

Cognitive AI for Real-World Systems

Agent-native cognitive systems that transform fragmented real-world data into reasoning, narratives, decisions, and trustworthy human-facing value.

This project applies agent-native infrastructure to transform fragmented real-world data into reasoning, narratives, decisions, and trustworthy value for human stakeholders.

Research Themes

Cognitive AI infrastructureIoT-to-narrative transformationDigital thread and real-world data integrationTrustworthy narrative generationDecision-support systems

Expected Artifacts

  • Cognitive AI testbed
  • IoT-to-narrative pipeline
  • Digital thread prototype
03Exploratory

Panic Engineering

Failure dynamics, control mechanisms, and reliability foundations of autonomous AI systems with real-world execution capabilities.

Panic Engineering is a research program focused on what happens when autonomous AI agents are given the ability to use tools, call APIs, execute commands, modify systems, and create real operational consequences. The goal is to study failure modes, control layers, observability, rollback, audit trails, and safety mechanisms for autonomous execution.

Research Themes

Tool-using agent failure modesAutonomous execution safetyPermission boundariesRollback and recoveryExecution tracingOperational riskHuman oversightIncident analysis for agentic systems

Expected Artifacts

  • Failure taxonomy
  • Panic Engineering technical report
  • Agent execution trace visualizer
  • Safety benchmark
Contribute

Interested in working on agent-native testbeds?

Join UsResearch Areas