Nathan Ben Soussan

I am an AI engineer who builds production-grade agent systems and scalable LLM inference infrastructure, prioritizing data sovereignty, low-latency inference, and cost-efficient retrieval. I design autonomous multi-agent workflows and RAG pipelines that improve developer velocity and system reliability.

My projects include architecting multi-agent facility management systems with local LLMs (Mistral 7B via Ollama) to cut cloud inference costs and guarantee privacy, engineering enterprise RAG solutions with LangChain and Pinecone, and optimizing local inference stacks with vLLM and quantization to increase throughput.

I combine deep DevOps experience with AI-first workflows—implementing queuing, rate-limiting, evaluation frameworks (LLM-as-a-Judge), and observability—to deliver resilient, high-performance AI platforms that reduce latency, lower costs, and accelerate feature delivery.

Available to work remotely from France.