At Navinspire IA, I pre-train and fine-tune open-source LLMs on NVIDIA H100 clusters, deploy distributed inference with Triton, and build production multi-agent and RAG systems. I also engineered AI backends for enterprise credit analysis, including prospection and scoring agents, explainable credit reports, and secure RAG chatbots.
I've delivered low-code AI workflow services, MCP integrations, vector-database agents, and enterprise controls across AWS, GCP, and Azure. My work spans the full AI pipeline, from data engineering and model training to scalable deployment, observability, and client proofs of concept.
