Skip to content
Holits

About

An independent platform engineer.

I'm Radu. I design, build and hand over production infrastructure on AWS, Kubernetes and Ethereum, after 20 years that started in carrier networks. The systems on this site are ones I have run, not slideware.

Mission

Make infrastructure boring again.

Most infrastructure problems are not exotic. They are the accumulated cost of decisions that were reasonable in isolation: a load balancer that cannot see the protocol it balances, a Terraform state that grew into a single blast radius, an observability stack that produces dashboards nobody opens.

My job is to make those systems predictable, legible and cheap to change, and then to leave, having documented why every part is the way it is.

Core expertise

What I am asked for

The disciplines the work actually falls into. The ones that are also service pages link through to how I approach them.

Platform Engineering
  • Cloud Architecture
  • Infrastructure as Code
  • CI/CD Engineering
  • Kubernetes Platforms
  • High Availability Systems
  • Performance Engineering
  • Infrastructure Scalability
  • Infrastructure Modernization
AI Infrastructure
  • LLM Infrastructure
  • AI Agent Infrastructure
  • GPU Infrastructure
  • AI Routing
  • AI Platform Engineering
  • Model Serving Infrastructure
  • MCP Infrastructure
Blockchain Infrastructure
  • Ethereum Infrastructure
  • RPC Infrastructure
  • Blockchain Node Operations
  • High-Performance RPC Routing
  • Archive Node Management
  • Blockchain Observability
Reliability
  • Site Reliability Engineering
  • Disaster Recovery
  • Capacity Planning
  • Incident Response
  • Observability Engineering
  • Monitoring & Alerting
  • Distributed Systems
Security
  • DevSecOps
  • Infrastructure Security
  • Supply Chain Security
  • Secrets Management
  • Compliance Automation
Networking
  • Traffic Engineering
  • Layer 4 / Layer 7 Load Balancing
  • Reverse Proxy Architecture
  • Zero-Downtime Deployments
  • Edge Infrastructure
Automation
  • Infrastructure Automation
  • Configuration Management
  • Workflow Automation
  • Release Automation
  • Platform Tooling

Philosophy

How I work

Part-time, async-first, with clear scope limits.

  • Part-time and async-first. Not an on-call service.
  • Fixed scope: each sprint has a price and a written deliverable, and the timeline is agreed on the scoping call, before it starts.
  • I reply to enquiries within two business days. On monthly support, the response time is the next business day.
  • Clear limits: hours do not roll over, and work beyond the agreed scope is quoted separately.
  • For larger Ethereum RPC and LLM routing work, I work with a second software engineer, who I have built these systems with.
  • I work inside your repositories and review process, and you own everything at the end.

Six principles that decide most arguments before they start

Engineering first

I am an engineer who consults, not a consultant who subcontracts engineering. The person who designs your system is the person who writes it.

Automation by default

If a change can be made by clicking, it will eventually be made by clicking at 3am by someone guessing. Infrastructure belongs in version control and is applied by machines.

Reliability over hype

A predictable system beats a fast one that sometimes isn't. Predictability is what everything downstream is built on, and it is usually what people mean when they ask for speed.

Production-grade quality

Health checks, metrics, failure handling and runbooks are part of the deliverable, not a follow-up phase that never gets scheduled.

Open-source mindset

I build on open source, contribute where I can, and write internal systems as though someone else will read them, because someone will.

Long-term maintainability

The measure of an engagement is whether your team can change the system confidently six months after I leave.

Career

From carrier networks to AWS.

  1. 2006

    Carrier networks

  2. Step 2

    Private cloud

  3. Step 3

    Azure and OCI

  4. Step 4

    AWS platform engineering

Employers: Oracle, Microsoft, Etisalat, Orange Romania, Huawei Enterprise, Adnet Telecom, GTS. These are experience, not clients.

Who is behind Holits

Written by a named engineer.

Radu Malica is a senior platform engineer with 20 years of experience across cloud platforms, virtualization, networking and distributed systems, in aviation, telecommunications, logistics and regulated enterprise environments. The work is modernising legacy platforms into ones that are reliable, observable and safe to release.

Day to day that means AWS, Azure and Oracle Cloud, infrastructure as code with Terraform and Terragrunt, Jenkins and GitHub Actions pipelines, Kubernetes and GitOps, and observability built on OpenTelemetry, Prometheus, Grafana, Loki and Tempo. Alongside client work I build independent projects in Labs: AI inference and LLM routing, Ethereum RPC routing and agent tooling. Several are described on this site, and Holits publishes Infrastructure MCP as open source.

Certifications

  • Google AI Essentials (Google)
  • Oracle Cloud Infrastructure 2020 Certified Architect Associate (Oracle)
  • AI Agents Fundamentals (Hugging Face)

The legal entity behind Holits is Holistic & IT Solutions S.R.L., VAT RO43533643, EUID ROONRC.J2021001957236.

Open source

What I give back

Infrastructure MCP is MIT-licensed and public: a self-hosted Model Context Protocol server that gives AI agents audited access to real infrastructure instead of raw SSH.

Other systems described on this site run on private infrastructure and are not published. I still document how they work, because the engineering reasoning is the transferable part.

Technology

What I actually use

Not a list of everything I have heard of. This is the set I work in regularly.

Cloud & virtualisation
  • AWS
  • Azure
  • OCI
  • OpenStack
  • VMware
Infrastructure as code
  • Terraform
  • Terragrunt
  • Ansible
  • Packer
Delivery
  • Jenkins
  • GitHub Actions
  • FluxCD
Runtime & traffic
  • Docker
  • k3s
  • ALB/ASG
  • HAProxy
  • NGINX
  • Apache
Observability
  • OpenTelemetry
  • Prometheus
  • Grafana
  • Loki
  • Tempo
Languages & systems
  • Go
  • Python
  • Bash
  • Linux

Send me your setup and I'll tell you what I'd check first.

Or email me directly at [email protected].