Younes Ben Tlili

I hold a dual role on Renault's Android Automotive platform (HMI): I designed the developer productivity AI layer (RAG + MCP) while leading the CI/CD platform and the team that runs it. It's the same CI/CD pipelines that industrialize AI — that's what makes it reliable and continuously maintained, not a POC.

AI & Developer Productivity

• Design and deployment of an internal RAG platform indexing source code repositories, exposed to developers via a custom MCP (Model Context Protocol) server integrated with GitHub Copilot — AI-powered code search and contextual assistance, directly in the IDE.

• Automated indexing pipelines (CI/CD) to ingest, vectorize (embeddings), and continuously refresh repos in the vector store → always up-to-date developer knowledge base.

• Reduced context-switching by exposing code, patterns, and documentation through natural language queries.

Platform & DevOps (the foundation)

• Leadership of the DevOps transformation and CI/CD chain for the HMI platform.

• Management of a System Team supporting 2 SAFe Agile Release Trains (15+ teams); management of a multidisciplinary delivery team of 8 people spread across France, Romania, and Tunisia.

• CI/CD chain optimization: Docker images, standardization and automation of CI jobs, release coordination and incident resolution; orchestration of a major CI migration impacting 15+ teams.

• Development of shared Kotlin plugins, adopted by all platform applications.

• Python automation (SAST report generation, project metrics collection and analysis) and enterprise-scale security tooling deployment (Checkmarx, SonarQube).

Technologies: RAG · MCP · vector database · embeddings · GitHub Copilot · Python · Docker · Kubernetes · GitLab CI/CD · Kotlin · Shell · YAML · SAFe

Founder of two automotive AI products that I designed and shipped end-to-end — from the agentic core to production on the stores. Hirrif is live on the App Store & Google Play. Concrete proof that I take an agentic system from concept to real production, with users.

AI & Agentic (the product core)

• RAG on a proprietary OBD-II diagnostic database (270+ fault codes) with hybrid search: DTC detection via regex combined with vector retrieval via Qdrant.

• Structured diagnostic analysis via the Claude API: JSON outputs, severity scoring, and actionable recommendations.

• Multi-agent workflows: pre-diagnostic, dispatch, and orchestration agents.

• Long-term data strategy for predictive maintenance and an automotive AI assistant.

Product & Architecture (proof it runs in production)

• Microservices backend (Node.js / TypeScript), real-time messaging, and AI service integration.

• Cross-platform mobile (Kotlin Multiplatform) and web dashboard (React / TypeScript).

• Multi-country SaaS architecture: configurable VAT, currencies, and payment providers; ZATCA e-invoicing, Tap Payments, WhatsApp OTP, and full Arabic RTL support for the Gulf market.

Technologies: RAG · hybrid search · multi-agents · LLM (Claude) · Qdrant · embeddings · Node.js / TypeScript · Kotlin Multiplatform · React · microservices · event-driven (RabbitMQ, Redis, MongoDB)

Rare niche: agentic AI applied to automotive diagnostics (OBD-II) — a high-value, underserved vertical.

Participation in the development of the new generation of Plato robots, within a team of 7 Android developers. Work on architecture, key features, and code quality.

Technical Stack: Android, Kotlin, MVVM, MVI, Mockito.