Hi, I'm Ahmed
Lead AI engineer who likes shipping ML systems end-to-end — LLMs, RAG, computer vision, and the infra that holds it all together.
My stack
About
Eight years in ML, mostly on the production side. I've shipped NLP systems for Arabic dialects, real-time computer vision for video proctoring, and — more recently — LLM and RAG pipelines that match candidates to jobs at scale.
The fun for me is going end-to-end: model research, training, serving, infra, and the orchestration glue. I work in Python, lean on KServe, Milvus, Temporal, and ArgoCD, and I'm happy writing CUDA-flavored Dockerfiles or pybind11 bridges when the problem calls for it.
Outside of work I write about what I'm learning, contribute to open source where I can, and try to keep the stack honest — fewer black boxes, more things I actually understand.
Experience
Skills
- LLMs / RAG / agentic systems
- Vision-Language Models (VLMs)
- Computer vision (YOLO, pose estimation)
- NLP (intent, entities, dialect ID)
- Python / FastAPI / Flask
- PyTorch / Transformers
- Self-hosted model serving (KServe)
- Vector search (Milvus)
- Workflow orchestration (Temporal, Airflow)
- GitOps deployments (ArgoCD)
- GStreamer / CUDA / pybind11
- Docker / Kubernetes