Anouar N.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Nov 5, 2026.

I build production-ready LLM applications, AI agents, MCP integrations, RAG systems, and custom fine-tuned models. I help teams move from AI prototypes to reliable systems with structured tool use, evaluation, APIs, and scalable GPU infrastructure. I’m an NVIDIA Deep Learning Institute Ambassador and NVIDIA Subject Matter Expert in Agentic AI, selected as one of 18 international SMEs supporting the preparation of NVIDIA Agentic AI certification content. I also work hands-on with large-scale NVIDIA GPU infrastructure spanning 116 GPU nodes: 79 A100 nodes, 35 H100 nodes, and 2 H200 nodes, totaling 928 GPUs for distributed AI, LLM fine-tuning, multi-node training, and HPC workloads. 🚀 What I can help you build: • AI agents and multi-step agentic workflows • Model Context Protocol (MCP) servers, clients, and tool integrations • Retrieval Augmented Generation (RAG), semantic search, and knowledge assistants • LLM fine-tuning with LoRA, QLoRA, PEFT, Hugging Face, and Unsloth • LangGraph, LangChain, and LlamaIndex orchestration • OpenAI, Anthropic Claude, and open-weight LLM integrations • FastAPI backends, REST APIs, Docker, and AI deployment • LLM evaluation, guardrails, observability, and human-in-the-loop workflows • Distributed PyTorch training with CUDA, NCCL, SLURM, Apptainer, and NVIDIA GPUs 🧠 Core AI & LLM expertise: Large Language Models (LLMs) · Generative AI · Agentic AI · AI Agent Development · Model Context Protocol (MCP) · Retrieval Augmented Generation (RAG) · LLM Fine-Tuning · Tool Calling · Function Calling · Structured Outputs · Context Engineering · LLM Evaluation · AI Safety · Prompt Injection Defense · Semantic Search · Vector Retrieval ⚙️ Engineering stack: Python · PyTorch · Hugging Face · LangGraph · LangChain · LlamaIndex · FastAPI · Docker · CUDA · NCCL · SLURM · Apptainer · SQL · Git · CI/CD Recent work includes: • OpsGuard AI: a secure agentic RAG and incident-triage system with controlled tool execution, policy enforcement, auditability, and reproducible security evaluation • Log Alchemist: an MCP-native agentic incident-response system for GPU/HPC environments • GPT-OSS 120B QLoRA: a distributed fine-tuning pipeline running across 24 NVIDIA H100 GPUs using Slurm, torchrun, NCCL, Unsloth, and LoRA 🎯 Why clients work with me: I work across the full AI stack, from model adaptation and distributed training to agent orchestration, retrieval, tool integration, evaluation, backend APIs, and deployment. I focus on systems that are reliable, inspectable, and engineered beyond basic chatbot prototypes. If you're building an AI agent, LLM application, RAG system, MCP integration, custom fine-tuning pipeline, or distributed AI workload, I can help take it from architecture and prototype to a robust engineering solution.