Athens, Greece · Open to collaboration

Konstantinos
Bardavouras

AI Solution Architect

Konstantinos Bardavouras

I design and ship enterprise-grade AI systems on Microsoft Azure — from Generative AI and RAG assistants to the secure landing zones and governance that make them production-ready. 15+ years of cloud, data and infrastructure engineering behind every architecture decision.

15+Years in enterprise IT
10+Certifications
4Azure & cloud expert roles
∞Curiosity for what's next
/ ABOUTagent.profile

From the datacenter floor
to the AI frontier

I'm an AI Solution Architect at PeopleCert, where I turn AI use cases from experiments into secure, scalable, reusable enterprise capabilities — knowledge assistants, AI agents and conversational experiences connected to trusted business content and workflows.

My path is unusual and it's my biggest asset: 15+ years across Unix systems, networking, enterprise storage, backup & disaster recovery, and hybrid cloud mean I understand the entire journey of a system — from the physical infrastructure to the tenant root to the LLM prompt. When I architect an AI platform, resilience, security and cost are not afterthoughts; they're in the blueprint from day one.

I thrive at the intersection of strategy, engineering and innovation — partnering with business and tech leaders to turn bold ideas into secure, data-driven outcomes.

packet→ platform→ prompt
/ APPROACHorchestrator

From AI initiative
to production

Every engagement runs like a multi-agent system: six specialised roles, one orchestrator — the architecture — and a human approval gate between every hand-off. Scroll to run the pipeline.

01
agent.discover

Discover

  • Use-case inventory
  • Risk classification · EU AI Act
  • Value & feasibility
02
agent.design

Design

  • Reference architecture
  • AI Landing Zone
  • Data access boundaries
03
agent.build

Build

  • RAG & agents · AI Foundry
  • Bot Framework
  • API Management gateway
04
agent.evaluate

Evaluate

  • Groundedness scoring
  • Red-teaming
  • Jailbreak & content filters
05
agent.ship

Ship

  • IaC · Bicep / Terraform
  • Private endpoints
  • CI/CD & runbooks
06
agent.operate

Operate & hand off

  • Monitoring & FinOps
  • Audit evidence
  • Ownership · least privilege
● HUMAN APPROVAL GATEbetween every stage·build → document → hand off
/ EXPERIENCEagent.history

Career timeline

Expand any role for the detail from my CV.

FEB 2026 — PRESENT

AI Solution Architect

PeopleCert · Athens
● CURRENT

Designing enterprise AI on Azure: Generative AI applications, RAG knowledge assistants, AI agents and bot interfaces over trusted content and secure APIs. Building the company's AI Landing Zone — identity, networking, guardrails, monitoring — and driving responsible AI governance aligned with the EU AI Act, ISO/IEC 42001 and NIST AI RMF.
  • AI Landing Zone for responsible AI adoption: subscription and resource structure, private connectivity, identity and access management, RBAC, managed identities, Key Vault, Azure Policy, logging, monitoring, secure deployment standards and guardrails.
  • Containers for custom AI services, Azure Functions for serverless automation, Bot Service / Bot Framework for conversational AI and API Management as the secure gateway for AI-enabled APIs.
  • Responsible AI: use-case inventory, risk classification, control mapping, audit evidence, automated evaluation, red-teaming, content filtering, groundedness validation and jailbreak protection with Purview, Defender for Cloud, Entra ID and Conditional Access.
  • Delivery approach — build, document, hand off: runbooks, RBAC maps, API flows and evidence, with ownership transitioned under least-privilege principles.
Azure AI FoundryAzure OpenAIRAGAI AgentsBot FrameworkAPI ManagementAI Governance
MAY 2025 — FEB 2026

Solutions Architect

Qualco Group · Athens

Architected secure, scalable Azure solutions with a focus on data platforms and AI-ready pipelines — from ingestion to deployment and observability — applying MLOps practices to bridge experimentation and production, with governance-by-design throughout.
  • Designing secure, scalable architectures using Azure PaaS components.
  • Building event-driven data pipelines and analytics foundations to support ML and AI workloads.
  • Applying MLOps best practices (deployment patterns, monitoring, traceability) to bridge experimentation and production.
  • Driving governance-by-design (Entra ID, RBAC, Key Vault patterns, data access boundaries).
Azure PaaSData PipelinesMLOpsGovernance
2012 — 2025 · 12+ YEARS

Cloud, Storage & Systems Engineering

OTE Group (Deutsche Telekom) · Athens

A 12-year progression through the full infrastructure stack: Application Support → Senior Unix Administrator (AIX, PowerHA, PowerVM) → Enterprise Storage & Backup (Dell EMC, IBM, Hitachi, HP, NetApp) → Cloud Engineer extending enterprise platforms to Azure with hybrid backup, DR and infrastructure-as-code.
02/2024 — 05/2025
Cloud Engineer

Hybrid backup and storage infrastructures extending on-premises platforms to Azure — Azure NetApp Files, custom azacsnap solutions, blob storage for backup; workload migration, IaC automation, identity and compliance for data protection.

01/2023 — 05/2025
Backup & Storage Cloud Engineer

Enterprise storage (SAN/NAS/Object) and backup platforms across multi-vendor environments; DR strategies aligned with business continuity; Commvault, NetBackup, NetWorker, DataDomain; automation of backup, restore and monitoring.

10/2020 — 10/2024
Enterprise Storage & Backup Administrator

Dell EMC (VMAX, PowerMax, VNX, Isilon, ECS), IBM (Spectrum, FS9000+, TSM), Hitachi (G1500, HNAS), HP (3PAR, Primera), NetApp; Brocade SAN zoning; custom monitoring with Zabbix, InfluxDB, Grafana and the ELK Stack.

10/2015 — 10/2020
Unix System Administrator

IBM AIX, Linux and Solaris in mission-critical infrastructures; PowerHA clusters for high availability and DR; PowerVM virtualization (LPARs, VIO servers, shared processor pools); shell automation.

09/2012 — 12/2015
Application Support Engineer

Stability, performance and availability of critical business applications at Cosmote — incident management, root cause analysis, release validation, integrations with CRM, billing and provisioning; SQL, Bash and Python automation.

Azure HybridBackup & DRSAN / NASPowerHA / PowerVMZabbix / Grafana / ELKAutomation
2011 — 2012

Early career

Android Developer · IT Engineer

Native Android development in Java (Google Play releases), network engineering and digital-transformation project evaluation — the foundations of a full-stack perspective.
04/2012 — 09/2012
Android Developer · Infokey Software Solutions

Native Java apps from concept to Google Play release — REST APIs, SQLite/Room, device sensors, push notifications, MVC/MVVM patterns.

08/2011 — 12/2011
IT Engineer · Globitel

Network set-up and device configuration (routers, switches); digital signage installation with ERP Pegasus, CRM and loyalty cards.

03/2011 — 07/2011
Investment project evaluation · Digital Aid S.A.

Action «Digi-Retail»: checking the correctness and technical training of investment plans.

Java / AndroidNetworking
/ SKILLSagent.skills

What I work with

Generative & Agentic AI

Production AI applications grounded in enterprise data — the core of what I do today.

Azure AI FoundryAzure OpenAIAI SearchRAGAI AgentsBot FrameworkAzure MLDatabricks

Azure Cloud Architecture

Landing zones, governance and cost-optimized reference designs.

Landing ZonesBicep / TerraformAKSContainer AppsFunctionsAPI Management

Security & AI Governance

Zero-trust platforms and responsible AI by design.

EU AI ActISO/IEC 42001NIST AI RMFDefenderSentinelEntra IDKey VaultPurview

Data & Analytics

Raw data to trusted insight — in hours, not months.

Microsoft FabricOneLakeSynapse DWReal-Time IntelligenceKQL

Resilience & Data Protection

Tiered backup, replication and disaster recovery at enterprise scale.

ASRCommvaultNetBackupRubrikZertoDataDomain

DevOps, Automation & FinOps

Pipelines, policy-as-code and continuous cost optimization.

GitHub ActionsAzure DevOpsAnsiblePython / BashCost Management
/ CREDENTIALSagent.credentials

Certifications

Azure Solutions Architect ExpertMicrosoft Certified · 2025
AI Transformation LeaderMicrosoft Certified · 2026
ITIL AI GovernancePeopleCert Certified · 2026
Azure Administrator AssociateMicrosoft Certified · 2024
Fabric Analytics EngineerMicrosoft · DP-600
Applied Skills: NLP with Azure AI LanguageMicrosoft · 2026
SC-100 Cybersecurity ArchitectMicrosoft · in progress
PRINCE2 AgilePeopleCert · in progress
Professional ML Engineer pathGoogle Cloud
Machine Learning with PythonCertification
CCNA SecurityCisco
HCIAHuawei Certified
Docker SecurityCertification
/ FIELD WORKagent.fieldwork

The day job, declassified

Classes of systems I architect in production — sketched without the confidential parts. No product names, no client data; just the shape of the work and the stack behind it.

SYS.01L10N // GENAI

Translation & Localization Pipelines

LLM translation at document scale — engineered for consistency, not one-off prompts.

  • ›Document pipelines: parsing, segmentation and batch orchestration over large content sets
  • ›Translation-memory reuse and terminology enforcement across every run
  • ›Automated quality scoring with human-in-the-loop review gates
STACKAzure OpenAIAzure FunctionsPythonTranslation Memory
SYS.02VOICE // CONVERSATIONAL

Conversational & Voice Agents

Assistants that hold a conversation over trusted content — in text, and out loud.

  • ›Bot interfaces wired to business workflows behind secure API layers
  • ›Voice loop: speech recognition → LLM reasoning → neural voice response
  • ›Guardrailed sessions: content filtering, context handling, escalation paths
STACKBot FrameworkAzure OpenAIAzure AI SpeechContainer Apps
SYS.03PLATFORM // GOVERNANCE

AI Gateway & Governance Automation

One controlled front door for AI workloads — and the evidence trail that proves it behaves.

  • ›Gateway patterns for LLM traffic: routing, quotas, content safety at the edge
  • ›Use-case inventory, risk classification and control mapping — automated
  • ›Red-teaming and evaluation pipelines producing audit-ready evidence
STACKAPI ManagementAzure PolicyPurviewEU AI ActISO/IEC 42001
SYS.04VISION // AERIAL

Aerial Search & Rescue Vision

Cloud-native computer vision for UAV search-and-rescue and flood response.

  • ›RGB and thermal video with GPS/IMU telemetry, streamed from the aircraft over MQTT/TLS into event ingestion
  • ›Frames land in a lakehouse and are scored by a YOLO model on a managed inference endpoint
  • ›Detections return as structured records and annotated frames — live operator overlay plus downstream reporting; validated end-to-end on an open aerial dataset
STACKComputer VisionEvent StreamingLakehouseManaged EndpointsEdge-ready
SYS.05FUSION // EMERGENCY

Emergency Intelligence Platform

Disaster-risk management for civil protection — many signals fused into one operational picture.

  • ›Emergency calls, IoT sensors, drone feeds, satellite imagery, incident history and social signals in one medallion lakehouse
  • ›ML pipelines for risk scoring; GIS and dashboards for situational awareness; alerting for field units
  • ›Governance, FinOps and business continuity documented as first-class deliverables — not afterthoughts
STACKData PlatformMLOpsGeospatialGovernancePublic Sector
/ PROJECTSagent.projects

Selected work

ENTERPRISE
  • L3 · Tenant root — management groups, policy inheritance
  • L2 · Subscriptions — identity, connectivity hub, AI workload spoke with private endpoints
  • L1 · Guardrails — Azure Policy, Defender, Sentinel, Log Analytics

AI Landing Zone

A secure foundation for responsible AI adoption at enterprise scale: subscription topology, private networking, identity & RBAC, policy guardrails, monitoring and secure deployment standards — so every AI workload starts compliant.

AzureSecurityGovernance
ENTERPRISE
  • L3 · Entry — user query through a secure API layer (APIM, Entra ID)
  • L2 · Pipeline — retrieve (AI Search, hybrid + vector) → ground on trusted docs with citations → generate (Azure OpenAI)
  • L1 · Safety — groundedness scoring, jailbreak shield, evaluation on every release

RAG Knowledge Assistants

Conversational AI and knowledge assistants grounded in trusted enterprise content — combining Azure OpenAI, AI Search and secure API layers, with automated evaluation, groundedness validation and jailbreak protection.

Azure OpenAIRAGResponsible AI
OPEN SOURCE ↗
$ azure-cost "chat assistant · 50k users/mo · EU residency"
SCENARIORELATIVEMONTHLY baseline€1,838 cost-optimized€1,318 high-performance€3,718
✓ 3 scenarios priced · 47 meters matched · trace → report.jsonl
ILLUSTRATIVE · SYNTHETIC DATA

CostPilot

Describe a workload in plain language, get a costed Azure architecture you can defend. An LLM plans the design; every price is deterministic — grounded on local snapshots of the Azure Retail Prices API, validated by per-service charge models and pricing invariants, and traceable down to the exact meter behind each number.

PythonAzureFinOpsMCP Server
IN DEVELOPMENT
  • L3 · Orchestrator — plans, routes and tracks cost across the team
  • L2 · Specialists — design, code, review and docs agents
  • L1 · Human gate — nothing ships without approval; live dashboard

Agentic AI Lab

A personal "AI software factory": multi-agent orchestration where specialized AI agents cover the full product lifecycle — design, code, review, docs — with human approval gates, cost tracking and a live dashboard.

Multi-agentOrchestrationLLMOps

Agora

PERSONAL PROJECT · IN DEVELOPMENT

An agentic analysis platform for the Athens Stock Exchange: a team of specialized agents reads the world's signals and turns them into a morning brief, an afternoon recap and event flashes — while a ledger scores every one of its own calls and feeds a weekly learning loop. Decision-support by design: no investment advice, no automated trading.

Multi-agentLLM PipelinesPythonDuckDBLocal-first

Also on GitHub

FabricExportAndBuild ↗Exports Microsoft Fabric metadata — workspaces, items, role assignments, capacities, domains — and renders it as readable console tables.PowerShell GradientDescentVisualization ↗A visual walkthrough of gradient descent — how an optimizer steps down a loss surface to find the parameters that minimize it.Python ChromiumBookmarksExport ↗Extracts only the bookmarks from Chrome and Edge across every profile, producing import-ready HTML plus JSON backups.PowerShell all repositories ↗

Selected writing

AI Governance in Seven Phases ↗Build the framework, document the operating model, hand accountability to named owners — in that order.FIELD NOTE · ON THIS SITE Networking Your Containers on Azure ↗A zero-to-hero guide to AKS kubenet and Azure CNI, Docker networking and Swarm overlay — in one cheatsheet.MEDIUM A Metadata Inventory Tool for Microsoft Fabric ↗Why keeping track of workspaces, items, roles, capacities and domains matters at scale — and what building the tool taught me.MEDIUM Storage Insights with NetApp IQ Unified Manager ↗A Python approach to pulling configuration insight out of enterprise storage estates.MEDIUM all articles ↗

EDUCATION

M.Sc. Health InformaticsNational & Kapodistrian University of Athens · 2007 — 2009
B.Sc. Computer ScienceAthens University of Economics & Business · 2002 — 2007
LanguagesGreek (native) · English (C1)

PUBLICATION

"Benefits and Restrictions in General and Thoracic Surgery"Zikos D., Bardavouras K., Mantas I. (2010) — Acta Informatica Medica, 18(3): 124–130
08 / CONTACT

Let's build what's next

Ready to accelerate your cloud and AI journey — or to co-create the next wave of AI-powered solutions? I'd love to hear from you.

✉ Email me LinkedIn GitHub Download CV ↓ PDF
ELSEWHERE Medium — writing DeviantArt — art