Cloud-native platforms
End-to-end platform design across application and infrastructure, with operability, security and local reproducibility built in.
AWS · Azure · Kubernetes
.NET · PostgreSQL · Helm
Solution architect · Utrecht, Netherlands
Achille
Tuglo
I turn complex systems into platforms teams can trust.
Cloud-native architecture, legacy modernisation and hands-on delivery for organisations where the software has to survive real business, security and production constraints.
Architecture becomes valuable
when it reaches production.
11+years building
production software
production software
3BrainBay engagements
in 12 months
in 12 months
2cloud ecosystems
AWS · Azure
AWS · Azure
CurrentlyConsultant at Xebia
What I do
From architecture decisions
to software that runs.
I work at the point where system design meets delivery. That means shaping the architecture, writing the critical code and leaving teams with systems they can operate and evolve.
Expertise
The hard parts,
made workable.
Legacy modernisation
Incremental routes out of monoliths and ageing services that protect existing callers, data and operational continuity.
Oracle PL/SQL · SQL Server
.NET Framework · WCF
Hands-on architecture
Architecture decisions grounded in implementation: APIs, domain models, event flows, deployment paths and the code behind them.
C# · TypeScript · SQL
DDD · CQRS · Event-driven
AI-assisted engineering
Deterministic guardrails around generated code, so teams can distinguish validator-clean output from software that is genuinely safe to run.
GitHub Copilot · Python
Kotlin · Validation systems
Selected work · 2017—2026
Evidence over
adjectives.
Selected engagements where the architecture had to answer to production.
Enterprise legacy modernisation, built around platform standards.
A reusable AI-assisted toolkit connecting Oracle analysis, reviewed specifications and Kotlin generation. It brings approved architecture guidance, application code, tests and delivery configuration into one workflow.
- Coverage
- 7
- target service patterns
- A common framework for APIs, background processing, event-driven services and more.
- Legacy estate analysed
- 1,810
- Oracle objects inventoried
- Across 3,434 source files in CWMS, with dependencies and migration boundaries surfaced for review.
- Documented local demonstration
- 5
- REST operations exercised
- A replenishment walkthrough reached local execution. Production integration decisions remained open.
Technical evidence
One toolkit. Seven service patterns.
- REST APIs
- Scheduled jobs
- Event consumers
- Event producers
- Workers
- Libraries
- Migration-only
Oracle analysis
Reviewed specifications
Kotlin generation
Verification
Built for the platform
Approved architecture guidance and reusable templates shape the output, including build configuration, Helm, tests and service documentation. A shared foundation for teams modernising different applications.
My contribution
In a two-engineer team, I owned generation and verification: reusable scaffolding, traceable model-authored logic and validators connecting generated outputs to their specifications.
Clear delivery decisions
Readiness reports turn unresolved architecture and integration questions into actions for engineering, QA and operations. Reviewers can see what has been checked and what still needs work before a release decision.
A valuation data platform, built with AI-assisted engineering.
I designed and built a greenfield platform for receiving, validating and distributing valuation reports, using AI-assisted engineering across application and infrastructure development. The delivery setup combines a reproducible local environment, automated tests and infrastructure-as-code with corporate deployment conventions.
Technical evidence
- Repeatable development
- Docker Compose and LocalStack provide a local environment for exercising submissions, storage and asynchronous processing. Smoke tests and automated test suites provide concrete checks for validating changes during AI-assisted development.
- Business value
- The platform turns XML and XBRL reports into validated, searchable valuation data. Automated schema and taxonomy registration reduces manual configuration, while provider-scoped access and domain-specific validation support controlled data exchange.
- Cloud delivery
- An API on AWS EKS connects to SQS-triggered Lambda processing, S3 and Aurora PostgreSQL. Separate application and infrastructure repositories define Helm and CloudFormation deployments, with build, test and analysis steps in Bitbucket Pipelines.
- Operational visibility
- OpenTelemetry traces follow processing across the API and worker. Dead-letter queues and CloudWatch alarms help surface failed or delayed processing for investigation.
AI-assisted engineering, grounded in real-world delivery.
I combine hands-on backend delivery with a project-specific AI engineering framework for Wonen's multi-repository .NET and React environment. Architecture conventions, business rules, ticket context and verification workflows support delivery efficiency, while engineers remain responsible for technical decisions and release readiness.
- Export formats
- 3XLSX · XLS · CSV
- Geographic index levels
- 5Local clusters to national
Technical evidence
- Repeatable workflow
- I configured reusable skills and specialist agent roles for analysis, implementation, testing and code review. Project-specific guidance captures existing patterns and lessons from previous reviews, helping subsequent changes build on that knowledge instead of repeatedly rediscovering it.
- Corporate standards
- AI-assisted changes are guided by existing architecture, shared libraries and API compatibility requirements. Build and test procedures, targeted business regression checks and structured PR reviews support verification before human approval.
- Delivery environment
- .NET 10 services target AWS EKS through Docker images and Helm deployments. Bitbucket Pipelines runs builds, tests and SonarCloud analysis, with a manual production deployment step. EF Core and PostgreSQL support persisted analysis data, alongside OpenSearch and a React frontend.
- Business delivery
- My backend contributions cover new-build pricing, geographic price-index selection and saved property comparisons. Persisted snapshots keep on-screen comparisons and exports aligned, while explicit pricing provenance distinguishes calculations based on individual homes from property-type estimates.
AI-assisted engineering for a connected property valuation platform.
I combined solution architecture and hands-on delivery with an AI-assisted engineering approach for Wizzer, a SaaS platform connecting homeowners, certified valuers and financial institutions in the Dutch residential property market.
Technical evidence
- AI-assisted delivery
- Applied the same principle later used on Wonen: grounding AI-assisted development in project context, existing architecture and business rules, with engineers retaining responsibility for technical decisions and validation.
- API alignment
- My contributions included generating TypeScript types and validators from OpenAPI, and refining API contracts to keep frontend requests and backend expectations aligned.
- Cloud-native platform
- A React and TypeScript frontend, delivered through S3 and CloudFront, connects to .NET 10 backend services with AWS Lambda processing. CloudFormation and Bitbucket pipelines support repeatable frontend deployment.
- Client continuity
- The first of three engagements NVM-BrainBay requested me for, across three different teams in twelve months, followed by Valuation Exchange and Wonen.
From stale batch reporting to live operational supervision.
Modernised a legacy estate built on .NET MVC, SQL Server 2008, SSAS, SSIS and large stored procedures. The rebuild pushed live state to the browser, replacing information that users had to request and wait for.
“The change that mattered wasn't the framework. It was moving supervision from something you requested to something that was simply true on the screen.”
Designing an Azure-native platform for external users.
Cloud-native development across the YOSS platform: microservices and Azure Functions, Service Bus integration, B2C authentication for external users and geolocation-aware search.
“The valuable part was the design work: proposing approaches to senior Microsoft architects, defending them, then turning those decisions into working software.”
A working principle
“An architecture decision that never gets implemented is just an opinion.”
So I stay close to the code.
Experience
A decade of systems
that had to hold up.
2025—Now
Xebia
Consultant · Netherlands
Three engagements at NVM-BrainBay: Wizzer, Valuation Exchange and Wonen. A fourth at Ahold Delhaize focused on AI-assisted legacy modernisation.
2018—2025
Independent
Software consultant
Six and a half years across industrial systems, e-commerce and legacy modernisation, including more than six years supporting EDF.
2016—2018
TALAN / ai3
Consultant
Cloud-native Azure platforms for Adecco and integration-heavy applications for Vinci Construction.
2014—2016
Sogeti
Consultant
Production support and evolution for Bouygues Telecom's pricing and online-sales applications.
Credentials
Certifications.
Industry credentials across the cloud and Microsoft application ecosystem.
Amazon Web Services
AWS Certified Developer - Associate
Microsoft
MCSD: App Builder
Education
Two schools.
One engineering path.
A joint-degree programme completed across Morocco and France.
2013—2014
Université de Bretagne Occidentale
MSc, Software and Embedded Systems
Brest, France
2012—2014
Institut National des Postes et Télécommunications (INPT)
Engineering Degree, Software Engineering
Rabat, Morocco
Capabilities
Tools follow
the problem.
No arbitrary percentages. These are the technologies and practices I use to move work from design to production.
Architecture
Solution Architecture · DDD · CQRS · Event-driven systems · REST · gRPC · ADRs · System Design
Engineering
C# · .NET · TypeScript · JavaScript · SQL · PowerShell · Python · Kotlin
Cloud & platform
AWS · Azure · Kubernetes · EKS · AKS · Lambda · Helm · Docker · CloudFormation
Data & operations
PostgreSQL · SQL Server · Oracle PL/SQL · Redis · OpenSearch · OpenTelemetry · CI/CD
Field Notes · Journal
Ideas tested
against production.
Validator-clean is not production-ready.
A practical framework for making AI-assisted modernisation reviewable without confusing validation evidence with release readiness.
About
Based in Utrecht
Working across Europe
Curious enough to challenge the design. Practical enough to ship it.
I'm a solution architect and engineer at Xebia in the Netherlands. Before that, I spent six and a half years as an independent consultant, mostly modernising the systems behind EDF's new-nuclear engineering supervision.
My best work sits between disciplines: understanding the business rule, drawing the system boundary, implementing the difficult path and making sure the result can be operated by the team that inherits it.
French · native
English · professional
Dutch · elementary
AWS & Microsoft certified
Start a conversation
05 / 05