Fabrizio De Cicco Résumé (PDF)

IoT Infrastructure Architect & Lead Engineer

Infrastructure,
by design.

I design cloud platforms for Culligan’s connected devices, and build the AI tools that help our engineers review, test and ship.

Explore the work Résumé (PDF)

Fabrizio De Cicco · Turin, Italy · Open to remote

Engineering assistant

Explore an agent
50 shared skillsTool integrations via MCP

Orchestrator

Receives a task, dispatches specialists and brings their findings together.

From the published case studyRead it ↗
50,000+devices in production

Architecture, reliability and engineering practice for Culligan's IoT platform.

10+years in infrastructure

~300,000fleet scale target

01 / Selected work

How a team works, built into its tools.

I designed and built a multi-agent engineering assistant for review, testing, release preparation and reliability. It is now used across the lead engineering team.

Explore the architecture

control

↓ dispatches to specialists

specialist agents

↑ all built on a shared foundation

shared

↕ connected to the toolchain via MCP

integrations · MCP

Source & PRsIssuesCI/CD Code qualityObservabilityChat
projectengineering-assistant
roledesign & build
built withClaude Code · MCP · TS
scale7 agents · 50 skills
statusin daily use
Orchestrator
routes work · plays the developer
  • Receives a task, dispatches the right specialists, assembles their findings into one verdict.
  • Acts as the developer-in-the-loop across the whole workflow.

Each agent owns one workflow and carries only the context it needs — organised by workflow, not by role.

Problem

Across a growing set of repositories, code review, QA, release coordination, and reliability monitoring were largely manual — slow to perform, hard to keep consistent, and a constant pull on senior engineers' time.

Approach

I designed and built a set of specialist AI agents in Claude Code, each owning a single workflow and sharing a common library of reusable skills. The orchestrator composes them into dynamic workflows that fan work out in parallel and adversarially verify findings before anything is surfaced. They integrate with the team's existing toolchain through MCP, sit behind safety guardrails that gate anything destructive, and the read-only ones run as scheduled loops that catch drift and post a summary — never acting unattended. The newest layer points the other way — hooks and skills that sharpen the engineer's side of the loop: reframing vague prompts at session start, surfacing blindspots before unfamiliar work, and quizzing the author on their own change before it ships.

four decisions that shaped it
  • Organise agents by workflow, not by role — so each carries only the context it needs.
  • Push shared conventions into a skills layer — so knowledge isn't duplicated across agents.
  • Make orchestration a workflow, not a script — fan out, verify, then synthesize.
  • Tier model capability by stage, like infrastructure — cheap executors for mechanical work, a top-tier advisor consulted on demand, and security-critical stages pinned so no cost tier can downgrade them.

Outcome

Rolled out to the lead engineering team and in daily use — seven agents sharing a 50-skill library, behind guardrails that gate anything destructive. Review quality has visibly improved and developer feedback has been consistently positive; the next iteration adds the numbers I want to manage it by: cycle time and defect leakage.

Explore the fan-out model

What parallel work changes

Modeled as a two-phase workflow: fan out → adversarially verify.

Modeled speedup 8.0× vs running them one after another

Equal task durations, separate capacity per stage and no coordination overhead. An illustrative model, not measured delivery time.

02 / Engineering practice

How I build

The decisions I come back to when building platforms and the teams behind them.

platform baselineShip identity, updates, and observability before a single feature.

For connected products, the platform's job is to keep a fleet secure, updatable, and visible — not to ship features on day one. Device identity, over-the-air update, and end-to-end telemetry are the load-bearing walls; bolt them on later and you're rebuilding the foundation under a live fleet.

first moveidentity + OTA + telemetry pipeline, before product features.
standardizationPick the boring defaults before the team is big enough to argue.

The cheapest time to standardize IaC modules, secret management, pipelines, and an SLO convention is at two repos, not twenty. Every week you wait, divergence compounds and migration turns political. Standardization isn't bureaucracy — it's the paved road that lets people stop reinventing plumbing.

first movea Terraform module library + one pipeline template + one observability convention.
kubernetesDon't run Kubernetes until the problem is Kubernetes-shaped.

AKS is right when you have many services, real scaling needs, and a team that can own the operational surface. For a handful of services it's a tax — managed container platforms ship the same outcome with a fraction of the ops. Complexity should be earned by load, not adopted for the résumé.

first movejustify the cluster against service count and on-call capacity.
reliability vs costAn SLO is a budget — spend it, don't gold-plate.

Chasing more nines than users actually feel just burns money and engineering time. Define the SLO from real user experience, then spend the error budget: ship faster while it's healthy, slow down and harden when it's not. Reliability and cost aren't opposites — the error budget is the dial that trades them on purpose.

first moveset SLOs from user-felt signals, then let the budget drive pace.
AI in deliveryLet agents propose; never let them push unattended.

AI agents are a force-multiplier on review, QA, and release prep — but anything destructive stays behind a human gate and a guardrail. Compose them into workflows that fan out and self-verify, give them a shared skills layer so conventions don't drift, and run the read-only ones on a schedule so problems surface on their own. The win is consistency and removed toil, not autonomy for its own sake.

first moveagents behind guardrails, gated on anything irreversible.
leadershipIf the team can't ship without me, I've failed.

A lead who hoards context becomes the bottleneck: the team stalls whenever they're unavailable, and burnout follows. My job is to make myself progressively unnecessary — mentor, document, pave the road — so the team moves faster than any one person could.

first movewrite it down, hand it over, and stay off the critical path.
03 / Experience

Where I've built things

  • Infrastructure Architect (prev. DevOps Engineer)

    Jul 2023 — present
    Culligan InternationalMilan · IoT · water
    • Lead IoT infrastructure architecture & reliability for the Culligan IoT Platform.
    • Own end-to-end infra design & operations — scalability, security, cost.
    • Mentor the DevOps/SRE team; evolve CI/CD, IaC & Datadog across global IoT.

  • DevOps Engineer

    Mar 2022 — Jul 2023
    MSC Mediterranean ShippingTurin · containers · scale
    • Built & maintained scalable CI/CD pipelines on Azure DevOps (YAML).
    • Automated provisioning with Ansible, Bicep & Rundeck — less manual config.
    • Stood up observability & logging with Datadog for faster detection.

  • Support Supervisor → Software Engineer

    Aug 2015 — Mar 2022
    DeltatreTurin · sports · real-time
    • Supervised a 12-person application support team.
    • Built backend & web apps (C#, SQL, ASP.NET, Angular) — Agile / TDD.
    • Kept microservices highly available on Azure with Docker & Kubernetes.
04 / The lab

The lab

Templates I actually use — de-identified and free to take — plus the number behind every "we need more nines" conversation.

Engineering artifacts

adr-template.md a decision, its context, and the trade-offs — so future-you knows why. Download
slo-definition.md SLI, target, error budget, and the burn-rate alerts that back it. Download
incident-review.md blameless postmortem: timeline, impact, root cause, actions with owners. Download
terraform-module-checklist.md the production-readiness gate before a module joins the paved road. Download
skill-template.md a Claude Code skill scaffold — trigger description, progressive disclosure, guardrails. Download

Each is a real .md in the /lab folder of this site's repo.

adr-template.md markdown · template 


        


        
Error-budget calculator

        
Pick an availability target and see how much downtime it actually buys you.

        

          

            

              
              %
            

            

              
              

                
                
                
                
                
              

            

          

          

          

            Error budget
            0.100%
            of the time you're allowed to be down
          


          

            
per day
1.44 min

            
per week
10.1 min

            
per month
43.2 min

            
per quarter
2.16 h

            
per year
8.76 h

          

        


        
Explore the command-line lab
        

          

            
            fabrizio / command-line lab
            interactive lab
          

          

            

            
          

        

        

        

      

    


    
    

      

        05 / Capabilities
        
Skills & certifications

        

          
Cloud & containers
Azure · AKS · Container Apps · Kubernetes · Docker · Helm

          
IoT & messaging
IoT Hub · Event Hub · MQTT · CoAP · LWM2M · device identity · OTA

          
IaC & CI/CD
Terraform · Terragrunt · Bicep · Ansible · Azure DevOps · YAML pipelines

          
Observability & SRE
Datadog · Grafana · Prometheus · SLOs · error budgets · incident response

          
AI-assisted delivery
Claude Code · MCP · multi-agent orchestration · model tiering · TypeScript

          
Certifications
CKA Certified Kubernetes Administrator (2022) · LFCS Linux Foundation Certified Sysadmin (2023) · AZ-104 Azure Administrator (2023) · AZ-400 DevOps Solutions (2023)

        

      

    


    
    

      

        06 / Get in touch
        
Let's build something reliable.

        
Architecture, platform reliability, or AI-assisted delivery — if any of that is on your plate, I'm happy to talk.

        

          Email
          GitHub
          LinkedIn
          Résumé (PDF)
        

        
based in Turin, Italy · open to remote