Gark Tech Contact
infrastructure · automation · applied AI

Where infrastructure meets intelligence.

Gark Tech is a learning lab exploring secure cloud infrastructure, automation and applied AI, and the foundation for a future solutions assistant.

secureby design, layered defence
automatedrepeatable, documented
intelligentAI where it adds value
// 01 approach

A reference architecture

The pattern the lab is built on: layered, least-privilege and recoverable. Shown as a design, not a live system map.

USERS

Secure access

Encrypted end to end, identity-first

→
EDGE

Protection layer

DNS, TLS, web application firewall, bot defence

→
GATEWAY

Reverse proxy

Routing, limits, hardened entry point

→
APPS

Containerised services

Isolated, reproducible workloads

→
DATA

Private data tier

Never exposed, least-privilege access

RESILIENCE

Backup & recovery

Off-site copies, regularly restore-tested

NEXT

Observability

Metrics, logs, traces and alerting in one view

FUTURE

AI operations

Models that read signals and assist engineers

// 02 ai & machine learning

Applied AI, grounded in operations

The AI track of the lab: practical machine learning for IT and business operations, always with a human in the loop.

llm/

Assistants & RAG

Language models answering from your own documents, runbooks and tickets, with sources cited.

ml/

Anomaly detection

Learning normal behaviour from network and server telemetry to flag problems early.

agents/

Workflow automation

AI agents that triage requests, draft responses and prepare changes for human approval.

data sources→clean & index→model / retrieval→evaluation→human review→deploy & monitor
// 03 stack

Tech stack

Solid: in use. Dashed: learning next.

net/

Networking

VLANsFirewallsWi-FiVPNSD-WANIPv6
sys/

Systems & Cloud

VirtualisationLinuxContainersBackup & DRKubernetesTerraform
sec/

Security

WAFTLS / PKIEDRIAMZero TrustSIEM
ai/

AI & Data

PythonAutomationLLMs & RAGVector DBsscikit-learnMLOps
// 04 projects

Lab projects

Each comes with a write-up: the problem, the design and the lessons learned.

~/lab/secure-cloudbuilt

Hardened cloud platform

A business application hosted behind layered edge protection.

  • Layered edge and origin protection
  • Least-privilege service identities
  • Automated, restore-tested backups
~/lab/service-deskbuilt

SLA-driven service desk

Requests routed and escalated automatically, with clear status for users.

  • Workload-based assignment
  • Timed multi-level escalation
  • Role-scoped workspaces
~/lab/ai-triageplanned

AI ticket triage

A model that classifies requests and suggests fixes from past resolutions.

  • Retrieval over solved tickets
  • Confidence scores, human approval
  • Measured against manual triage
// 05 roadmap

Where the lab is heading

From a single hardened server to automated, observable, zero-trust infrastructure.

phase 01 · done

Foundation

  • ✓Cloud host and domain
  • ✓Edge TLS and WAF
  • ✓Off-site backups
phase 02 · done

Business platform

  • ✓ERP and service desk
  • ✓SLA automation
  • ✓Role-based access
phase 03 · next

Observe & automate

  • ○Metrics and dashboards
  • ○Infrastructure as code
  • ○CI/CD pipeline
phase 04 · future

Zero trust & AI

  • ○No inbound ports
  • ○SIEM and alerting
  • ○AI ops assistant
● future plan · in design
// 06 vision

Gark Tech Solution Assistant

The long-term goal of this lab: an AI-powered assistant that helps teams plan, secure and run their technology, combining infrastructure know-how with machine learning.

A concept in development. Not a commercial offering.

Our branch Wi-Fi keeps dropping in the afternoons. Where do I start?
Likely causes from similar cases: channel congestion at peak hours, or DHCP pool exhaustion.

Suggested checks
1 · Client count vs pool size
2 · Channel utilisation by hour
3 · AP event log around drop times
Ask about networks, servers, security…⏎
// contact

Follow the build.

Questions about the lab, or want to swap notes on infrastructure and AI? Get in touch.

✉ [email protected]