Aymane Nouhail

Spearheaded the V2 ground-up rewrite of the company's agentic AI platform as technical lead, driving architecture decisions and coordinating development across creative, media, and PR teams.

• • Architected a production multi-tenant agentic platform serving 4 enterprise clients across real-time collaboration, docu ment editing, and conversational AI, processing 300K+ monthly requests through a shared provider-agnostic LLM layer supporting Claude (Anthropic), Gemini (Google), and GPT (OpenAI).

• • Designed a modular agent architecture with clear orchestration boundaries between planning, execution, and tool layers, enabling new client workflows to ship without modifying core platform logic.

• • Built an evaluation and observability stack combining agent testing frameworks with Langfuse tracing, Application Insights, and structured logging to monitor agent behavior, latency, and failure modes in production.

• • Reduced campaign production time by 30% by co-building a RAG pipeline for Carrefour radio-ad generation (500+ historical campaigns ingested, LLM-drafted scripts with brand-tone consistency).

Built an end-to-end voice-first agentic assistant on Raspberry Pi, combining wake-word detection, automatic speech recog nition (ASR), and a retrieval-augmented LLM pipeline for hands-free smart-home control and knowledge-base Q&A.

• • Designed a streaming speech-to-text pipeline with wake-word detection feeding a context-aware LLM inference layer, enabling fully hands-free natural-language control of IoT devices.

• • Implemented a RAG agent grounded on a structured knowledge base, supporting device automation commands, multi media queries, and open-domain Q&A with low-latency on-device inference.