Networks that stay secure, reliable and automated.

L3NET brings independent senior engineers to firewall estates, data-centre networks and third-line support. Directly with you, or through a prime.

We take a whole project, or one role in yours. Remote or at your site.

These figures are the specialists' experience. They are not all contracts held by L3NET OÜ.

Services

What we do

We build new networks and keep existing ones running. Bring in one senior specialist for a single task, or several for a whole programme, from design to third-line support.

Design

Architecture

Designs that show how your sites, data centres and cloud connect, and what happens when a link or device fails. For a new network, or as a design review of the one you run. You get documents and diagrams your team can build from.

Low-level designs and migration plans for data centre and firewall moves, in the semiconductor sector.

See the semiconductor work

Network security

Firewalls, edge to core

Rules nobody can explain, and nobody dares to remove. We design firewall policy and run the firewalls day to day. We also handle rule review and audit, and migration from one vendor to another, at the internet edge and inside the network.

A utility perimeter moved onto next-generation firewalls, with intrusion prevention and malware sandboxing.

See the perimeter rebuild

Data security

Data protection and AI security

Staff use AI tools that nobody approved, often called shadow AI. Firewall and proxy logs can show which tools are in use, by whom, and how much is uploaded. We allow the approved tools one user group at a time. DLP controls reduce the risk of sensitive data leaving by email, web upload or removable storage. We also assess local, self-hosted LLM options for clients who cannot send their data to a public AI service.

We offer hands-on AI and LLM training for your team, remote or at your site.

Not included: legal advice, EU AI Act or ISO/IEC 42001 audits, and AI model testing.

DLP for a utility, upgraded and tuned for email, web upload and removable storage.

A control on the use of AI tools at work, set up and now in use.

Ask about AI tools at work

Data centre

Data centre

We read your current estate before we touch it. Then we plan, build and run the switching, designed so that one failed switch or link does not stop your services. Old equipment comes out in phases.

Data centre services for a telecom operator, moved onto Cisco ACI with load balancer integration.

See the telecom work

Application delivery

Load balancing and CDN

Application delivery with load balancers, certificates, custom traffic rules and CDN set-up. Your applications stay up and fast.

Lost client addresses behind a load balancer were traced with packet captures. The header is to be trusted only when the load balancer, or an origin allowlist, can see that the request came from the CDN. That change is still planned.

See the consolidation work

Automation

Network automation and cloud

Automation that asks first. Network changes are written as code from one source of truth, and a person approves each change before it goes live. Drift checks report differences and never fix on their own. Cloud networks follow the same rules.

Proof of concept, 2026: a self-hosted network automation platform, built so a business can move between cloud providers more easily. It has one source of truth, sends every change through a reviewed merge request, and needs no internet access at run time.

Ask about automation

Operations

Support and escalation

Senior third-line support for your team, or for your own customers. We find the cause of hard faults and prove it with data before we fix it. Remote or at your site, in normal hours. If a contract needs 24/7 escalation, we add people to cover it.

Main escalation point, with root-cause analysis, for a data centre and firewall estate for more than four years.

Ask about third-line cover

Platforms we work with

Specialists who partner with L3NET have worked hands-on with more than 15 vendors. These are the main ones. L3NET does not resell hardware, so the advice is not tied to a brand.

  • Cisco
  • Palo Alto
  • Check Point
  • Fortinet
  • F5
  • A10
  • Akamai
  • Google Cloud
  • Oracle Cloud
  • Arista
  • Juniper
  • Nokia

Product names belong to their owners. They show experience, not a partner status.

How we differ

The common problems

In our experience, projects go wrong on order, knowledge and paperwork more often than on technology. Here is what we do about each.

Common project problems, what L3NET does about each, and the proof
The problem, as you say itWhat we doProof
“It is a black box. Nobody knows what is connected to what.”We read the network before we change it. Read-only commands collect configs, port use and optics. Then we draw the real picture. Designs, procedures and flow maps are part of the job, and you own them.Read-only collection scripts for configs, ports and optics, on a current data centre project. Design blueprints and application flow maps for a data centre move.
“Only the supplier knows how it runs.”We hand over in working sessions as well as documents. Your engineers learn the system before we leave.Knowledge transfer for an in-house team on three platforms.
“Tickets go round in circles, and the same fault comes back.”Hard issues go to a senior engineer at third line. Where we can, we prove the cause with packet captures or live data before we propose a fix.Main escalation point, with root-cause analysis, for a data centre and firewall estate. Cause of lost client addresses behind a load balancer proved with packet captures.
“Every change is a gamble.”Every step has a check before, a check after and a rollback. We test in a lower environment first, or say plainly that there is none. We roll out in phases: log first, enforce second.Scripted pre-checks and post-checks, pass or fail per step, on provider link maintenance. A client-address fix planned in two phases: log first, then block.
“The people who sold it are not the people who do it.”Once we assign an engineer to your project, that engineer does the work. If the job needs more people, we name them before it starts.In the examples below, a senior engineer, architect or lead did the work. Those engagements were not all L3NET contracts.

When an engineer is away

A named specialist stays on the job. A dedicated specialist works on that project only. If they are away, and the contract allows, someone who knows the work covers. The documents are written so that person can pick it up. For a larger scope, we add more people.

Selected work

Before and after

These examples are from regulated and security-sensitive sectors, so the organisations are not named. They show work specialists who partner with L3NET have done. They were not all L3NET contracts. In each one, a senior engineer, architect or lead did the work. References from former contacts are available on request.

Semiconductor sector

Data centre and firewall overhaul

Before
A legacy data centre network. Firewall rule bases split across two vendors.

What was done
Low-level designs and migration plans were written for a move to Cisco ACI. The firewall rule bases were converted from both vendors to Palo Alto, and checked before cut-over. Custom F5 traffic rules were built for per-user access to development environments, and tested before production.

After
One firewall vendor instead of two, and per-user access to development environments.

Cisco · Palo Alto · F5

Proves Architecture, Firewalls

Telecommunications

Network and security operations for a telecom operator

Before
The data centre network was old and still had to move to a new fabric. The in-house team needed training on the new fabric, the firewalls and the load balancer. Documentation was missing.

What was done
Data centre services were moved onto the new fabric, joined up with the load balancer and the virtual server platform. Technical operations for the fabric and the firewalls were led for more than four years, as the main escalation point, looking for the root cause of problems, not only the symptom. Design blueprints and application flow maps were written, and the in-house team was trained on the data centre fabric, the firewalls and the load balancer.

After
Data centre services were moved to the new fabric. The design and the application flows were written down. The in-house team was trained to run the new platforms.

Cisco · Palo Alto · F5 · Check Point

Proves Data centre, Support and escalation

Energy

Perimeter security rebuild for a utility

Before
An ageing perimeter: separate firewalls, a web proxy, a legacy intrusion detection system and an older DLP product.

What was done
The perimeter was moved onto next-generation firewalls, with intrusion prevention and malware sandboxing. Web proxy services were moved onto the firewalls. DLP was upgraded and tuned for email, web upload and removable storage.

After
The separate web proxy and the legacy intrusion detection system were retired. The firewalls now do the proxying and the intrusion prevention.

Palo Alto · Fortinet

Proves Firewalls, Data protection and AI security

Defence

Multi-pod fabric operations in a high-security environment

Before
A growing multi-pod, multi-tenant data centre fabric, with monitoring that no longer fitted it.

What was done
The operations team was led. New pods and sites were designed and built, with new fabric monitoring. The documentation and the migration plans were reviewed.

After
New pods and sites were put into service, with monitoring that matched the fabric.

Cisco · Palo Alto

Proves Data centre

Online platformsIn progress

Edge and data centre consolidation

Before
The data centre core had reached end of life. The upgrade path was blocked: the new software was not compatible, and a known bug could crash the core during an upgrade. The load balancer did not pass on the client address, so the applications logged the load balancer's address instead. Traffic that skipped the CDN was still accepted. Old equipment for a service that had moved out still used rack space.

What was done
The upgrade and replacement options were set side by side, with the risk and the hardware for each. The cause of the lost client addresses was proved with packet captures, before any change was proposed. The fix was planned in two phases: log traffic that skips the CDN first, then block it. Trust client-address headers only when they come from the CDN. The clear-out of the old equipment was planned in two phases: first the equipment of the departed service, then the shared equipment.

After
The assessment is done. A proposal to move to a new switching platform, with risk-reduction steps first, is with the client. Engineering and day-to-day support continue. Traffic that skips the CDN is planned to be logged first, then blocked. The client-address header is to be trusted only from the CDN. The clear-out is planned and will free two racks.

Cisco · F5 · Akamai

Proves Data centre, Load balancing and CDN

Public sector

Endpoint and web DLP for a public agency

Before
An endpoint DLP product that was being replaced.

What was done
Endpoint DLP was moved to a new product, set up to find sensitive data on desktops. DLP for web and social-media traffic was designed and tuned. Test evidence was produced with the security team and the penetration testers.

After
DLP was in place on desktops and web traffic, with test evidence produced with the security team and the penetration testers.

Proves Data protection and AI security

How we work

Four steps, then you own it

A delivery project has four steps. Your team then owns it. A specialist who joins your team follows your process.

L3NET can contract directly with end clients, or work through a prime contractor or integrator, including in public-sector and defence procurement.

  1. Step 1

    Assess

    We map what you have, how traffic really flows, and where the risk sits. Read-only first.

  2. Step 2

    Design

    A low-level design and a plan with a rollback for every step.

  3. Step 3

    Build

    Changes in agreed maintenance windows, tested before and after each step.

  4. Step 4

    Hand over

    Runbooks, design documents and knowledge transfer. Your team owns it when we leave.

Two ways to work with us

A group for your project, or one senior specialist in a role inside your team. How many people depends on the project.

  • A group for your project. The work runs from assessment to handover, with one senior specialist or several.
  • A specialist in your team. A named specialist joins your team for a set period and works with your processes and tools.

Roles we take

  • Project and programme management. Plans, risks, suppliers and reporting for a larger programme. Availability on request.
  • Architect. Designs the network and its security, and writes them down.
  • Network security consultant. A senior security expert who joins your team for a set period, under your processes and your security management system.
  • Embedded engineer. Works inside your team, or inside an integrator's team for its end customer.
  • Third-line support. Root-cause analysis for critical incidents, remote or at your site. We add people for 24/7 escalation if the contract needs it.

Certifications and clearances, confirmed per project

Specialists who partner with L3NET have held senior security certifications, and vendor certifications on the platforms they work on. L3NET confirms the current status for each project. Named certificates are not listed here. TODO: confirm which certifications may be named.

Security clearances belong to people, not the company. For each contract L3NET proposes people who hold the clearance it needs, or who can be put forward for vetting. The type of clearance is confirmed then, not published here.

They know the rules of regulated and high-security sites. L3NET answers supplier due-diligence questions on request.

Network

Specialists

A trusted network of independent senior engineers. L3NET brings them in as each project needs. They partner with L3NET OÜ. They are not employees, and their past projects were not all L3NET contracts. Some have worked together for about seven years, others for about five.

The name

From layer 3 to the whole path

L3NET began as Layer 3 Networks: the routes between sites. The same care now runs from the first survey to the application. A project can need one part of that path, or all of it.

Talk to an engineer

Layer 8: you. Every network problem eventually reaches this layer. Let's talk.

Send us the fault, the plan or the tender. We will tell you which parts we can cover, and with how many people.

Or message the founder on LinkedIn: Egidijus Stankevičius, opens in a new tab.

TODO The company LinkedIn page is not confirmed. The link above is his own profile.

A few lines are enough. No passwords or configs, please.

We use your name, email and message to reply to you. Read our privacy notice.