10/14/2026 updated
Premium member
100 % availableSenior AI Engineer & Architect | Agentic AI, LLM, MCP | Java / Kotlin, Kubernetes, Azure | 10+ yrs
Saint-Gratien, France
Only remote
Diplôme d'ingénieur in Software Engineering (Master's level), EMSI Rabat, 2015About me
I turn vague business goals into systems that run in production, and I stay accountable for them afterwards. AI only holds up there on a solid backend and clean delivery: that combination is what I bring, from architecture to the last pipeline. English and French, based in France.
RAGGitHub ActionsOpenAIAzure DevOpsLangChainJava (Programming Language)Application Programming Interfaces (APIs)Artificial IntelligenceMicrosoft AzureContinuous DeliveryContinuous IntegrationPostgreSQLSoftware ArchitectureAnsibleSystems Integration
More than ten years of software engineering, the last seven in banking and insurance (BNP Paribas CIB, AXA). Four years of LLM systems in production, agentic systems since 2023, built on Java / Kotlin backends, Kubernetes and Azure.
Selected results
- AXA: re-architected an actuarial platform used across several countries, replacing background synchronisation jobs with an event-driven design on Azure Service Bus that carries four times the previous compute load.
- AXA: shipped a generative AI assistant for C++ code completion used by actuarial teams.
- BNP Paribas CIB: as technical lead of six to eight people, built a self-service orchestrator that provisions Jenkins, Artifactory, SonarQube, Argo CD and other DevOps tools through the bank's IT marketplace, cutting turnaround from weeks to hours.
- Neurisense, my company: AI in production for clients, including a GenAI recommendation engine that helped a cosmetics e-commerce site increase sales, and AI-driven sourcing of construction materials.
AI engineering
- Agentic AI and multi-agent systems in production; MCP servers exposing business APIs to agents as tools (TypeScript / NestJS)
- RAG with OpenAI and PostgreSQL vector search, tool calling, LangChain
- OpenAI, Anthropic Claude and Azure OpenAI APIs; LLM observability, cost and usage tracing in Langfuse
- Token economics: token budgets per feature, context selection, caching, model choice per task
- AI-assisted engineering: coding agents daily for three years (Codex, OpenCode, Claude Code); workshops for developer teams on AI coding
Backend and architecture
- Java / Kotlin (10 years), Vert.x, Spring Boot, Spring Security, Hibernate / JPA
- TypeScript / Node.js, NestJS, React, Next.js, Go, Python scripting
- Event-driven architecture, microservices, contract-first REST APIs (OpenAPI), OAuth2 / OIDC, multi-tenant PostgreSQL
Cloud and DevOps
- Kubernetes (8 years), Docker, Helm; Microsoft Azure (4 years): Azure Batch, Azure Service Bus, Azure DevOps
- CI/CD with Jenkins, GitHub Actions and Azure DevOps; Ansible
English and French. Based in France, working remotely, with occasional on-site visits.
Experience
Professional experience10+ years
Freelancer since2019
Languages
ArabicNative speakerEnglishNative speakerFrenchNative speaker
Project history
AI agents and LLM features built into client businesses; architect and developer from scoping to go-live.
- MCP servers exposing each client's business APIs to agents as tools on two production projects: TypeScript / NestJS, explicit input and output schemas, traces and costs in Langfuse.
- Retail: GenAI recommendation engine for a cosmetics e-commerce site, which helped increase sales.
- Construction: AI-driven automation of construction-materials sourcing for a construction company.
- RAG and tool-using agents with OpenAI and Anthropic models, PostgreSQL as vector store.
- Inference cost control: token budget per feature, context selection, model choice per task, usage traced in Langfuse.
- Workshops and training material for developer teams on AI coding.
Technologies: OpenAI, Anthropic Claude, Claude Code, MCP, LangChain, Langfuse, PostgreSQL, TypeScript, NestJS, Docker
Azure platform for actuarial risk management and financial projections, used across several AXA countries; end-to-end owner of its backend, from architecture to operations.
- Re-architected the platform: replaced background synchronisation jobs with an event-driven architecture on Azure Service Bus that distributes load across instances, handling four times the previous compute load.
- Designed the distributed execution of C++ actuarial projections on Azure Batch, with Service Bus messaging between compute and application services.
- Java / Kotlin and Vert.x services on PostgreSQL, multi-tenant by country.
- Shipped to production a generative AI assistant for C++ code completion used by actuarial teams.
- Containerised services delivered to production on Kubernetes through Azure DevOps pipelines.
Technologies: Java / Kotlin, Vert.x, PostgreSQL, Azure Batch, Azure Service Bus, Kubernetes, Docker, Azure DevOps, generative AI
Self-service provisioning orchestrator for DevOps tools, published through the bank's internal IT marketplace; owned end to end, from architecture to run.
- Led a team of six to eight people.
- Automated nearly all DevOps tool provisioning (Jenkins, Artifactory, Confluence, Jira, Bitbucket, Argo CD, Argo Workflows, SonarQube): turnaround cut from weeks to hours.
- Java / Kotlin and Go services, React / TypeScript interfaces, Python scripting, on Docker and Kubernetes.
Technologies: Java / Kotlin, Go, React, TypeScript, Python, Docker, Kubernetes, Jenkins, Artifactory, Argo CD
Certificates
SAP Hybris Commerce Developer 6.2
SAP2018
SAP Hybris Product Support Specialist 6.0
SAP2018
SAP Hybris Business Analyst 6.3
SAP2018