.png&w=256&q=75)
1
1
Estonia
6+ years of experience
DevOps, Cloud Security and Agentic systems enginineering specialist based in Estonia, designing intelligent infrastructure that bridges LLM orchestration, autonomous agents, and enterprise-grade cloud deployments. I specialize in building agentic AI pipelines, RAG-powered workflows, and managed cloud platforms — from fine-tuned models to production-ready IaaS. Every system I ship is designed for reliability, scale, and real-world automation.

nullstate is a local-first purple-team CLI for infrastructure-as-code security validation, built for the AMD x lablab.ai hackathon. It takes Terraform projects, runs them against local sandbox targets, detects exploitable cloud storage misconfigurations, uses a self-hosted model endpoint for red-team and blue-team reasoning, applies deterministic Terraform remediation, reruns validation, and generates reviewable evidence artifacts. The final demo validates two scenarios: an AWS S3 public access block misconfiguration and an Azure Blob container with anonymous access. In both cases, nullstate shows the attack path succeeding before remediation, applies a Terraform patch, and verifies that the path is blocked afterward. Each run produces report.md, findings.json, events.jsonl, metrics.json, attack.py, and remediation.patch. The project uses Python, Typer, Rich, Terraform, LocalStack, vLLM, ROCm, and AMD MI300X-hosted model inference through an OpenAI-compatible endpoint. The core security verdict is deterministic for reliability, while the model adds adversarial reasoning, remediation explanation, and case-study-quality reporting.
10 May 2026