Kiran Krishna — Melbourne-based senior AI engineer and AWS architect with 14+ years across government, transport, rail, AWS cloud platforms, and agentic systems. Currently leading agentic AI delivery at the Department of Transport and Planning Victoria.
An enterprise agentic AI platform: a reusable, governed agent runtime that separates the agent layer from domain skills, so business areas automate document-heavy, governance-sensitive workflows on shared infrastructure — with orchestration, traceability, guardrails, and human oversight built in.
◆ Reusable runtime for org-wide automationWhere I test agent patterns end to end. Jarvis OS — a local-first personal AI operating system routing a primary agent to specialists, with memory and approval-gated tools. AgentCouncil — an open-source Claude Code MCP plugin for structured multi-agent deliberation.
★ Open source · multi-agent runtime testbedA secure, multi-account AWS foundation on Control Tower and Landing Zone Accelerator — centralized networking and auditing, standardized guardrails and governance, SSO + CyberArk via IAM Identity Center, and SOC/SIEM security monitoring — giving teams a governed, compliant platform to build on.
▲ Secure, compliant multi-account foundationCloud-native industrial-IoT applications on AWS for rolling-stock and locomotive health: real-time, fleet-wide health monitoring, an image ingestion & search system for fault analysis, and streaming sensor analytics across edge and cloud — serving global rail operators.
▲ Fleet-wide operational visibilityA free platform I built to teach building AI agents — from fundamentals to the frontier. Structured courses across agentic AI, Claude Code, Codex, AI leadership, and Amazon Bedrock AgentCore — read or listen, with slide decks, hands-on labs, and quizzes. No signup, no paywall.
I'm Kiran Krishna, based in Melbourne. I've spent 14+ years building enterprise software, AWS cloud platforms, and AI workflows across consulting, transport, rail, and government — an AWS Certified Solutions Architect & AI Practitioner.
My current work is production-grade agentic AI, built mostly on AWS: systems that use tools, delegate to specialist agents, preserve audit trails, respect approval boundaries, and operate inside real enterprise constraints. At DTP VictoriaI lead agentic AI delivery and designed reusable agent-runtime patterns for business-wide automation. Before that I architected cloud-native AWS systems for transport & rail platforms at GE/Wabtec, serving global operators across many sites.
Outside work I build Jarvis OS, AgentCouncil, and personal AI research systems. That matters because I don't treat AI as a slideware topic — I build with it every day, test patterns in real workflows, and bring the useful lessons back into enterprise delivery.
Working on an agentic system, an AI platform, or enterprise AI delivery? Drop a note — I read everything.