At Cyware Labs, I build and own production agentic AI platforms, including MCP runtime infrastructure, distributed LangGraph orchestration, multi-agent systems, and enterprise RAG pipelines.
I've shipped a FastMCP server over SSE for multi-tenant enterprise tools, extended MCP client capabilities for dynamic tool loading, and built a LangGraph agent engine with checkpointing, async interrupt/resume, and cross-restart idempotency. I also architected a Go and Python platform using NATS JetStream and PostgreSQL-backed state.
My work automated Tier-1 SOC alert triage, reducing analyst MTTR from about 25 minutes to 8 minutes. I also helped deliver a no-code AI workflow builder that reduced build time from one hour to two minutes and redesigned backend infrastructure to cut costs by 60% at 10× throughput.
Outside my core platform work, I'm building persistent memory and security tooling for AI agents, including semantic retrieval, MCP tools, NLI-based grounding verification, PII detection, prompt-injection classification, and schema-drift alerts.
