
1
1
India
1 year of experience

AgentTriage automates SRE incident triage using a three-agent LangGraph pipeline running on AMD Developer Cloud. When production breaks, engineers waste hours reading logs manually. Our system fixes that. The Planner agent reads logs and service metrics to form a triage strategy. The Executor takes step-by-step actions — classifying severity, identifying root cause, sending remediation commands, and resolving the incident. The Summarizer generates a structured post-incident report. Built on AMD MI300X with 192GB VRAM running Qwen2.5-72B via vLLM. Deployed on HuggingFace Spaces with a live interactive demo. Achieved perfect scores across all three incident scenarios.
10 May 2026