AI Engineer specializing in Production LLM Systems, RAG & Agentic AI

I design and deploy evaluation-driven AI systems with a focus on reliability, observability, inference efficiency, and measurable quality. My background combines Generative AI research, Computer Vision, and large-scale model optimization, enabling me to bridge the gap between research and production.

🎯 Currently focused on RAG / LLM engineering and agentic AI.

Evaluation-driven retrieval system built on FinanceBench.

Key features:

• Multi-provider LLM abstraction (LiteLLM)
• Agentic workflows (LangGraph)
• Hybrid retrieval with Qdrant
• Automated evaluation pipeline (Ragas)
• Full observability (Langfuse)
• High-performance inference benchmarking (vLLM)
• Document ingestion with Docling

Focus areas: retrieval quality, hallucination reduction, observability, latency optimization, and reproducible evaluation.

• LLM & RAG — LangChain · LlamaIndex · LangGraph · LiteLLM · Qdrant · Docling · Ragas · Langfuse · vLLM · Ollama · Groq
• Generative AI & CV — PyTorch · Hugging Face (Transformers, Diffusers, timm) · diffusion & flow matching
• Serving & infra — FastAPI · Docker · AWS · SLURM · Singularity
• Eval & observability — Ragas · Langfuse
• Core — Python · Linux

• Generative AI — motion forecasting, and image/video synthesis & editing
• Computer Vision — learning-based image & video compression and anomaly segmentation

• 📧 Email — ghorbel.ahmd@gmail.com
• 💼 LinkedIn — ahmed-ghorbel97