Panagiotis Mavrikos is joining NOFire AI as a Member of Technical Staff on the Core Infrastructure team. He joins from Nubificus, the systems company founded by NOFire AI co-founder and Chief Scientist Anastasios Nanos, where he spent three years working on serverless and lightweight virtualization and contributing to urunc, the CNCF sandbox project that lets OCI tooling run workloads inside lightweight virtualized environments.
He steps into the seat where safe autonomous action stops being a design goal and becomes infrastructure that has to run, reproducibly, inside real customer environments. Making agent execution isolated, observable, and safe enough for production is a genuinely hard systems problem, and it is the one Panagiotis came here to work on.
Background
At Nubificus, Panagiotis worked close to the metal on how workloads are packaged, isolated, and run. He specialized in serverless computing and lightweight virtualization, and became fluent across the cloud-native stack: Kubernetes, Karmada, Docker, and the tooling around them.
There he contributed to urunc, the sandbox project created by Charalampos Mainas, now also on the team at NOFire AI, and worked on European Horizon research projects along the way. He has also written about the systems he works in, from extending the Kubernetes API with custom resources to running serverless workloads on Kubernetes with Knative.
Alongside that work he co-founded Omnyo and is pursuing a Master's degree focused on software development and artificial intelligence. First place at the AI Hackathon Greece is one of a few awards he has picked up building things fast with other people.
The through-line is systems that have to be correct under real constraints: packaging, isolation, orchestration, and the runtime boundary itself. That is exactly the vantage point the Core Infrastructure team at NOFire AI needs.
What Panagiotis will own
As part of the Core Infrastructure team, Panagiotis owns the foundations that let NOFire AI run reliably across environments: cloud infrastructure, Kubernetes, deployment automation, and the sandbox and customer environments where agents actually execute. A big part of that is bringing urunc into the platform, so agent workloads run inside lightweight, strongly isolated environments by default.
The goal is to make agent execution isolated, reproducible, observable, and safe enough for production use. It is the layer everything else at NOFire AI stands on, and the one customers should never have to think about.
More about Panagiotis
How did you get into systems work?
I started by wanting to understand what happens beneath the application layer: how software is deployed, how services communicate, and what keeps a system reliable in production. That curiosity began during my studies in Digital Systems, where I was introduced to operating systems, networking, cloud computing, and software engineering, and gradually led me into Kubernetes, distributed infrastructure, and open-source systems work. I am also currently pursuing a Master's degree focused on software development and artificial intelligence. Through my previous work I contributed to European Horizon research projects and to urunc, a CNCF sandbox project that enables OCI tools to run workloads inside lightweight virtualized environments. Those experiences made it clear that systems engineering was the area where I wanted to build my career.
Why NOFire AI?
NOFire AI is working at the intersection of the areas I care about most: infrastructure, distributed systems, production reliability, and AI. What stood out to me is that the company is not treating safe autonomous action as a prompting problem. It is building the execution boundaries, identity controls, and production context required for agents to act safely in real environments. That is a difficult systems problem, and exactly the kind of problem I want to work on.
What will you be working on here?
I am part of the Core Infrastructure team, working on the foundations that allow NOFire AI to run reliably across environments. My work includes cloud infrastructure, Kubernetes, deployment automation, sandbox and customer environments, and the integration of urunc into the platform. The goal is to make agent execution isolated, reproducible, observable, and safe enough for production use.
Tools you rely on?
Linux and Kubernetes are at the center of most of my work, alongside Terraform, Argo CD, Docker, Helm, GitHub Actions, and AI development tools. Together they help me build, automate, and operate reliable infrastructure efficiently.
Outside of work?
I am naturally curious and spend a lot of my free time learning new technologies or building side projects to understand how things work. I enjoy hackathons and have been part of teams that won awards, including first place at the AI Hackathon Greece. I also enjoy reading history, playing sports, and exploring ideas outside my immediate area of work.
Panagiotis is based in Greece, and joins a systems team building the infrastructure that makes safe agent execution real.
