Souhila Dehni

Project: Development of an intelligent voice bot for Leroy Merlin's customer service, based on Agentic, internal API, and generative AI. Development of a generative AI voice bot for Leroy Merlin's customer service. NLU + GenAI for intent detection and automated responses. Implementation of RAG and integration via backend API (orders, GDPR, identification). Conversational orchestration and fallback management to a human agent. Tech Stack: Python, GCP (DataStore, GeminiEnterpriseForCustomerExperience (Cxagentstudio), Conversation Agents, GCS), Github, API (FastAPI), Agentic, skills=[]

Project: API Agent / Improve and automate the API lifecycle via Generative AI. Definition of the architecture and collaboration with the solutions architect: design and implement the scenario. Implementation and industrialization of a RAG preparation agent: Design and deployment of a data pipeline to extract APIs from the Service Registry and transform them via GenAI and LangGraph. Deployment on Cloud Run, orchestrated by Cloud Scheduler with support for full and delta loading modes (daily update of modified APIs only). Industrialization of a Swagger generation agent: Development of a FastAPI API deployed on Cloud Run and orchestrated with LangGraph, allowing the generation of OpenAPI/Swagger files from invocations to the internal GenAI platform (Next Gen AI). Collaboration with DevOps teams for setting up CI/CD pipelines and automating deployments in the production environment. Tech Stack: Python, GCP (cloudrun, cloudstorage, cloudscheduler, vertex), FastAPI, LangGraph, Docker, Gitlab, GitlabCI/CD, skills=[]

Project: Load testing for scaling from 500 agents to 5000 agents. Deployment of the chatbot for 5000 agents. Opening up to a wide panel of agents. Scoping, development, and industrialization of a monitoring dashboard on Datadog, and implementation of metrics to monitor the chatbot's performance evaluation. Perform the RUN process (bug fixing and feeding a backlog of functional evolutions), VSR (regular service verification), and preparation for TMA (Third-Party Application Maintenance). Tech Stack: Python, Langchain, FastAPI, AWS, Azure, RAG, OpenAI, GPTo-mini, Docker, shell, Gitlab, Datadog (APM, DDTrace, LLMObsevability, Auto-instrumenter), Locust, RAGAS, Monitoring/Logs, Openlens, Langfuse., skills=[]