Konstantinos Chaitas kchaitas

Konstantinos Chaitas, a portrait on a dark background, in a checked shirt

(Software & Cloud Architect)

Multi-tenant SaaSAWSAI-driven engineering

  • 20+ years
  • AWS Certified Cloud Practitioner
  • Available for projects
CertificationAWS Certified Cloud PractitionerMay 2026

I design multi-tenant SaaS platforms and build the AWS infrastructure they run on myself. 20+ years in technology: from networks to SaaS, and now AI agent systems that write and review production code.

Work

since May 2026Live

Praion

Co-founder

AI-first solutions for businesses. We help companies use artificial intelligence where it pays off: in the visibility of their answers (AEO), in how they talk to their customers, and in custom solutions built around their own needs.

  • AEO
  • AI-first
  • Custom solutions
Visit praion.gr
Praion: three fields around one companyDiagram: Praion in the centre, with three fields around it — AEO, customer communication and custom solutions.AEOCustomer communicationCustom solutionsPraionAI-first solutions for businessesPraion: three fields around one companyDiagram: Praion in the centre, with three fields around it — AEO, customer communication and custom solutions.PraionAEOCustomer communicationCustom solutionsAI-first solutions for businesses

since February 2026Live

Praion AI

Co-founder & Chief Architect

A SaaS platform for marketing and communication with AI, made of three applications:

  • Leads Concentrator — one inbox for every communication channel, with an AI assistant.
  • Product Manager — SEO and AEO for websites and online shops, with our own WordPress plugin and Shopify app.
  • Content Studio — marketing content produced and published with AI.

I own the architecture and lead the team technically. I built Content Studio myself, along with the billing system for AI usage, the Autopilot for autonomous replies and the architecture of the WordPress plugin. I connected several AI providers — Claude, Gemini, OpenAI — and generative media services: Imagen, Veo, Runway. I led the security review, with automated access-control checks in CI.

  • Multi-tenant SaaS
  • AI
  • Laravel
  • Next.js
  • Marketing
Visit praion.gr
Praion AI: one core, three applicationsDiagram: six AI providers (Claude, Gemini, OpenAI, Imagen, Veo, Runway) connect from above to the platform core, which feeds three applications: Leads Concentrator, Product Manager and Content Studio.AI providersClaudeGeminiOpenAIImagenVeoRunwayPlatform coremulti-tenant · AI usage billing01Leads Concentratorinbox + AI assistant02Product ManagerSEO/AEOWP · Shopify03Content Studiocontent with AI1 core · 3 applications · AI providers plugged in from outsidePraion AI: one core, three applicationsDiagram: six AI providers (Claude, Gemini, OpenAI, Imagen, Veo, Runway) connect from above to the platform core, which feeds three applications: Leads Concentrator, Product Manager and Content Studio.AI providersClaudeGeminiOpenAIImagenVeoRunwayPlatform coremulti-tenant · AI usage billing01Leads Concentratorinbox + AI assistant02Product ManagerSEO/AEO · WP · Shopify03Content Studiocontent with AI1 core · 3 applications ·AI providers plugged in from outside

Nicosia · since 2019Live

iwebforyou

Founder & CEO

My own company. We build websites, online shops and platforms on our own headless CMS, and we also take care of the infrastructure they run on. I work directly with founders and executives: from the first conversation about what they need to a site that is running and maintained.

  • Custom websites
  • Platforms
  • SaaS
  • Cloud
Visit iwebforyou.com
iwebforyou: four fields around one companyDiagram: iwebforyou in the centre, with four fields around it — custom websites, platforms, SaaS and cloud.Custom websitesPlatformsSaaSCloudiwebforyouNicosia · 2019from design to delivery and operationiwebforyou: four fields around one companyDiagram: iwebforyou in the centre, with four fields around it — custom websites, platforms, SaaS and cloud.iwebforyouNicosia · 2019Custom websitesPlatformsSaaSCloudfrom design todelivery and operation

2020 — 2026No name and no link, on purpose

Healthcare SaaS platform in America

Software Architect / Chief Technology Architect

A SaaS platform in America for quality and compliance in healthcare. It is multi-tenant, with a separate database for each customer, and has modules for analytics, standards, policy management and inventory.

I was its architect, with the final say on every technical decision. Six years, in production.

I designed, built and operated the AWS infrastructure on my own.

  • Multi-tenant
  • Healthcare compliance
  • AWS
  • Database per tenant
A separate database for each customerDiagram: one application layer, below it the tenant router, and below that N isolated databases — one per customer, with dashed walls between them.One application layerTenant routerDB ACustomer ADB BCustomer BDB CCustomer CDB …NCustomer …NN isolated databases · one customer, one databaseA separate database for each customerDiagram: one application layer, below it the tenant router, and below that N isolated databases — one per customer, with dashed walls between them.One application layerTenant routerDB ACustomer ADB CCustomer CDB BCustomer BDB …NCustomer …NN isolated databases ·one customer, one database

Next.js

Headless CMS & site factory

In-house product

The CMS under every website and online shop we deliver. It covers online shops, bookings and multilingual content; SEO and AEO; GDPR compliance; payments through Stripe, PayPal and Viva; and product feeds for Skroutz, BestPrice, Shopflix and Google Merchant.

A new site is generated automatically from the base, and upgrades reach every running site in a controlled way, not by hand one site at a time. We keep adding features, and each one becomes available to every site.

I designed it. A developer on my team builds it.

  • Next.js
  • E-commerce
  • SEO/AEO
  • GDPR
  • Feeds
The site factory: one base, N sitesDiagram: one headless CMS base on the left. Upgrades pass through a single checkpoint and reach every site (Site 1, Site 2, Site N) on the right in a controlled way.One CMS baseheadless · Next.jscontrolledupgradeSite 1Site 2Site …None base → N sitesupgrades arrive in a controlled way, not one by oneThe site factory: one base, N sitesDiagram: one headless CMS base on the left. Upgrades pass through a single checkpoint and reach every site (Site 1, Site 2, Site N) on the right in a controlled way.One CMS baseheadless · Next.jscontrolledupgradeSite 1Site 2Site …None base → N sitesupgrades arrive in a controlled way,not one by one

Claude Code

Agent Studio & Planner

In-house tools · in production

Planner takes a specification and delivers a working application. On its own it goes through architecture, implementation with tests written before the code, quality checks, an audit and deployment. It saves its state at every step, so if it stops it resumes where it left off. One project runs on one machine at a time, with a lock and with memory and load limits so tasks never collide. The owner gives feedback in rounds: they write what they want, the system applies it and shows the result again. Human approval is needed only at deployment.

Agent Studio is a set of specialised agents on Claude Code that run in parallel: code review, security audit, pentest, QA exploration, UX and SEO/AEO. Each one looks at its own area and writes up its findings.

What keeps them useful is not the model. It is the mechanisms. Completion criteria that leave no room for interpretation. Every finding checked against the current code before it is accepted. A final check that is never delivered red. Automatic rollback if a deployment fails. And every question answered once becomes a rule, so it is not proposed again.

Both are used in production, including for the security review of Praion AI

  • Claude Code
  • Multi-agent
  • TDD
  • Code review
  • Security
The Planner loopDiagram: specification, architecture, tests before the code, implementation, check, deploy. Three brakes: completion criteria, verification of every finding, automatic rollback. Human approval is needed only at deploy.01Specification02Architecture03Testsbefore the code04Implementation05Check06Deploybrake 1Completioncriteriabrake 2Verification ofevery findingbrake 3Automaticrollbackhuman approval only herespecification → working applicationThe Planner loopDiagram: specification, architecture, tests before the code, implementation, check, deploy. Three brakes: completion criteria, verification of every finding, automatic rollback. Human approval is needed only at deploy.01Specification02Architecture03Testsbefore the code04Implementationbrake 1Completioncriteria05Checkbrake 2Verification ofevery finding06Deploybrake 3Automaticrollbackhuman approval only herespecification → working application

What I do

Platform design

From a blank page to production: how data is split per customer, where one service ends and the next begins, what becomes a queue and what becomes an API. I make the technical decisions and stay responsible for whether they hold.

AWS infrastructure & security

Designing, building and running production infrastructure. Network, databases, secrets, backups, logging, threat detection, identities. And the daily watch over all of it.

AI-driven engineering

AI agent systems that write and review code — with completion criteria, verification of every finding before it is accepted, and automatic rollback when something fails.

Products from zero

From an idea to a working product: technical strategy, team, implementation, delivery and operation. Working directly with founders and executives.

(ABOUT ME)

20+ years, one thread

I started as an electronic and network engineer: Intracom first, then ANCO, with a CCNA and a CCNP. In 2009 I founded Entermac and ran it for ten years — projects inside complex client environments, contracts in regulated sectors and critical infrastructure, and the job of keeping systems working year after year.

Since 2019 I have run my own company, iwebforyou, and it is still going. Since May 2026 I have also been a co-founder of Praion. Alongside, I spent six years as architect of a healthcare platform in America.

When I design a platform I do not stop at the diagram. I go down to the network, the database, the backups and who has access to what. A system is judged in production.

Konstantinos Chaitas · Software & Cloud Architect

Konstantinos Chaitas, seated at his desk with a laptop in front of a large window

The tools I work with every day

  1. AWS

    17 services in one production infrastructure that I designed and built alone.

  2. Laravel

    The backend of the platforms: REST APIs, queues, WebSockets.

  3. Next.js

    The frontend, and the headless CMS behind every site we deliver.

  4. PostgreSQL

    The data model, and the separation of customers in multi-tenant SaaS.

  5. Claude Code

    The base of Agent Studio and Planner.

  6. GitHub Actions

    Automated access-control checks in CI.

(NUMBERS)

A system is judged in production

  • Backend PHP · Laravel · REST APIs · PostgreSQL
  • Frontend React · Next.js · TypeScript · Tailwind
  • Cloud & DevOps AWS · Docker · GitHub Actions · Linux
  • AI Claude Code · multi-agent · LLM
  • Quality & security TDD · application security

(AWS)

I built the infrastructure single-handedly

For the healthcare platform in America, the production AWS infrastructure was my work from start to finish: I designed it, I built it and I ran it, with no infrastructure team behind me.

CertificationAWS Certified Cloud PractitionerMay 2026
The AWS infrastructure in six layersThe 17 AWS services sorted into six layers: network (VPC, Route 53), compute (Elastic Beanstalk), data (RDS, S3, AWS Backup), identities and secrets (IAM Identity Center, Secrets Manager), monitoring (CloudTrail, Config, CloudWatch) and defence (GuardDuty, Security Hub, WAF, and my own system with Lambda, EventBridge, DynamoDB).01NetworkVPCRoute 5302ComputeElastic Beanstalk03DataRDSS3AWS Backup04Identities & secretsIAM Identity CenterSecrets Manager05MonitoringCloudTrailConfigCloudWatch06DefenceGuardDutySecurity HubWAFLambdaEventBridgeDynamoDB17 services · 6 layers · one infrastructure, built by one personin lime: my own serverless defence systemThe AWS infrastructure in six layersThe 17 AWS services sorted into six layers: network (VPC, Route 53), compute (Elastic Beanstalk), data (RDS, S3, AWS Backup), identities and secrets (IAM Identity Center, Secrets Manager), monitoring (CloudTrail, Config, CloudWatch) and defence (GuardDuty, Security Hub, WAF, and my own system with Lambda, EventBridge, DynamoDB).01NetworkVPCRoute 5302ComputeElastic Beanstalk03DataRDSS3AWS Backup04Identities & secretsIAM Identity CenterSecrets Manager05MonitoringCloudTrailConfigCloudWatch06DefenceGuardDutySecurity HubWAFLambdaEventBridgeDynamoDB17 services · 6 layers · oneinfrastructure, built by one personin lime: my own serverlessdefence system
Every service and what it does
  • Elastic Beanstalkruns the application
  • VPCnetwork and isolation
  • RDSdatabases
  • S3file storage
  • Route 53DNS
  • Secrets Managersecrets and keys
  • CloudTraila record of every action
  • Configconfiguration tracking
  • GuardDutythreat detection
  • Security Hubsecurity findings in one place
  • IAM Identity Centeridentities and access
  • AWS Backupbackups
  • CloudWatchdaily watch over logs and security
  • WAFapplication protection
  • Lambda · EventBridge · DynamoDBmy own serverless defence system

A system that learns who is knocking

On top of AWS WAF I wrote my own serverless system with Lambda, EventBridge and DynamoDB. It does not rely on lists of known-bad addresses. It spots malicious scanners by how they behave and blocks them automatically.

Catching a scanner: list or behaviourTwo paths side by side. Left: a list of known-bad addresses, which only knows what has already been recorded and lets a new scanner through. Right: the behaviour pattern is recognised, passes through Lambda, EventBridge and DynamoDB, becomes a rule and the scanner is blocked automatically at the WAF.Requests knocking at the doorList of known addressesknows only what is already recordeda new scanner walks through the doorvsBehaviour → rule → blockthe behaviour pattern is recognisedLambdaEventBridgeDynamoDBrule → blocked automatically, by the WAFCatching a scanner: list or behaviourTwo paths side by side. Left: a list of known-bad addresses, which only knows what has already been recorded and lets a new scanner through. Right: the behaviour pattern is recognised, passes through Lambda, EventBridge and DynamoDB, becomes a rule and the scanner is blocked automatically at the WAF.Requests knocking at the doorList of known addressesknows only what isalready recordeda new scanner walks throughthe doorvsBehaviour → rule → blockthe behaviour patternis recognisedLambdaEventBridgeDynamoDBrule → blocked automatically,by the WAF

Path

Ten years running my own technology company: projects inside complex client environments, contracts in regulated sectors and critical infrastructure, and responsibility for long-term operation.

The path from 2005 to todaySix bars over time. Closed, with a start and an end: 2005–2006 Intracom, 2006–2009 ANCO, 2009–2019 Entermac, 2020–2026 healthcare platform in America. Open, reaching today: iwebforyou since 2019 and Praion since 2026.20052010201520202025Intracomelectronic engineer2005 — 2006ANCOnetwork engineer2006 — 2009Entermacmy own company2009 — 2019iwebforyouFounder & CEO2019 → todayHealthcare platformChief Technology Architect2020 — 2026PraionCo-founder2026 → todayThe path from 2005 to todaySix bars over time. Closed, with a start and an end: 2005–2006 Intracom, 2006–2009 ANCO, 2009–2019 Entermac, 2020–2026 healthcare platform in America. Open, reaching today: iwebforyou since 2019 and Praion since 2026.20052010201520202025Intracomelectronic engineer2005 — 2006ANCOnetwork engineer2006 — 2009Entermacmy own company2009 — 2019iwebforyouFounder & CEO2019 → todayHealthcare platformChief TechnologyArchitect2020 — 2026PraionCo-founder2026 → todaytoday
  1. 2026 — todayCo-founder, Praion — AI-first solutions for businesses
  2. 2019 — todayFounder & CEO, iwebforyou Ltd, Nicosia
  3. 2020 — 2026Chief Technology Architect, healthcare SaaS platform in America
  4. 2009 — 2019Founder & CEO, Entermac, Kozani / Heraklion
  5. 2006 — 2009Network Engineer, ANCO SA, Athens
  6. 2005 — 2006Electronic Engineer, Intracom SA, Athens

Certifications

  • AWS Certified Cloud PractitionerMay 2026
  • CCNPCisco, 2008
  • CCNACisco, 2006
  • HubSpotJuly 2020

Education

  • MSc Computing SystemsUniversity of Macedonia, Thessaloniki, 2011–2013
  • BEng Electronic EngineeringTechnological Education Institute of Athens, 1999–2005

Languages

  • Greek — native
  • English — professional
  • Italian — professional
  • German — learning

Writing

(HOBBY · MOUNTAINEERING)

In the mountains, out of signal

When I am not designing systems, I am climbing. Olympus, Kaimaktsalan, Vovousa to Valia Calda: on foot, over the last few years.

Mountaineering is closer to my work than it looks. You check the weather before you set out. You plan the route around what you can handle, not what you wish you could handle. You pack what you will need if something goes wrong. And you turn back when the data says so, not when stubbornness does.

Up there the phone has no signal. It is the one place where no system needs me.

Konstantinos Chaitas with a backpack and trekking poles among pine trees, Olympus behind him
Olympus, 2024
Konstantinos Chaitas in a gorge, in front of a rock face covered in grass
May 2022

Let's talk about your next system

If you have a platform to design, an infrastructure that has to hold, or a team that needs technical direction — write to me.

Greece — remote, across Europe and the US

VIDEO CALL

Book a video call

Pick a free time from my calendar. You will get a Microsoft Teams meeting invitation by email.

Security check — loads when you start filling in the form.

* Required field

  • Multi-tenant SaaS
  • AWS
  • Laravel
  • Next.js
  • AI agents
  • Application security
  • TDD
  • PostgreSQL
  • React
  • Systems architecture