I build ML systems that are technically solid, reproducible, and easy to evaluate — clear problem framing, structured pipelines, measurable results.
yousef = {
"role": "MLOps Engineer | ML Platform Engineer",
"background": "5+ years in Data Engineering — ETL/ELT, Data Warehousing, Python, SQL",
"platform": ["Kubernetes", "Kubeflow", "Istio", "Katib", "PyTorch"],
"ml_focus": ["NLP", "Computer Vision", "Reinforcement Learning", "ML Systems"],
"approach": "research → reproducible implementation → production-ready pipelines",
"location": "Italy 🇮🇹",
}| Area | Tools |
|---|---|
| ML & AI | Python PyTorch Transformers QLoRA Scikit-learn |
| Domains | NLP Computer Vision Remote Sensing Multimodal RL |
| Systems | Kubernetes Docker Istio Katib kind Git |
| Frontend | Vue 3 Pinia Tailwind CSS |
|
Unified NLP experimentation for sarcasm & sentiment classification across model families and English varieties.
📊 |
🔥 WildFireMultimodal geospatial CV predicting burned area from pre-fire satellite, terrain, weather & infrastructure data.
📈 IoU |
🤸 RLMuJoCo Hopper robust locomotion transfer under dynamics shift — PPO + curriculum domain randomization + entropy scheduling.
📈 |
⚙️ ML_OpsPrivate-cloud-style ML workflow on a local Kubernetes cluster — experiment infrastructure, deployment runbooks, browser IDE.
🏗️ kind · MetalLB · Katib · end-to-end local ML infrastructure |
Best entry points for reviewers: NLP · Computer Vision · RL · MLOps