Machine Learning Engineer · Data Scientist
Computer vision · medical & health ML · genomics · NLP / RAG — and production engineering to ship it.
📍 Casablanca, Morocco · 🌍 English & French · 🔗 nizarassad.github.io
I build machine-learning systems end to end — from the model to the service that serves it. My current focus is LLM / RAG / retrieval systems, agents & automation, and MLOps — taking models out of the notebook and into production (clean packages, tests, containers, CI). I also have a deep foundation in computer vision, health & medical ML, and genomics.
- 🤖 LLM & retrieval engineering: RAG, embeddings, semantic search, evaluation harnesses
- ⚙️ ML engineering / MLOps: reproducible pipelines, model serving (FastAPI), Docker, CI
- 🔬 Applied ML foundation: deep learning for detection & classification, medical prediction, cancer genomics
- 🛒 I also ship production software (Shopify apps & integrations) — proof I deliver, not just prototype
🤖 LLM · MLOps · Data Engineering — current focus
| Project | What it does | Stack |
|---|---|---|
| rag-semantic-search | Offline-first, production-grade RAG: pluggable embeddings/index/LLM, FastAPI, Docker, honest eval harness (real Recall@k / MRR) | Python, scikit-learn, FastAPI |
| mlops-pipeline | End-to-end MLOps: train → track (MLflow) → version → serve (FastAPI) → drift-monitor (PSI). Reproducible, tested, containerized | XGBoost, MLflow, FastAPI, Docker |
| llm-eval-kit | Test LLM outputs like code: deterministic + semantic + grounding + LLM-as-judge scorers, precision/recall vs gold, pytest-style assertions | Python, scikit-learn, pydantic |
| ecommerce-data-pipeline | ELT into a DuckDB warehouse: staging→marts star schema, data-quality tests, DAG orchestration (Prefect + fallback) | DuckDB, SQL, Prefect, pandas |
🔬 Machine learning & data science
| Project | What it does | Stack |
|---|---|---|
| Weapon-Detection | Real-time weapon detection to help address gun violence | YOLOv5, PyTorch, OpenCV |
| Pan-Cancer-Analysis | Genomic analysis on The Cancer Genome Atlas (TCGA) | Python, bioinformatics, ML |
| Stroke-prediction | Predicting long-term stroke risk from clinical data | scikit-learn, ML |
| Heart-disease-detection-system | Likelihood of heart disease from patient features | scikit-learn |
| Face-mask-detection | Real-time mask detection & classification | CNN, OpenCV |
| Digits-recognition | MNIST classification comparing SVM vs CNN | scikit-learn, deep learning |
Production engineering
| Project | What it does | Stack |
|---|---|---|
| shopify-carrier-service-demo | Live checkout shipping rates with every failure mode handled | Node.js, Express, REST |
| shopify-theme-update-safe | Customizations that survive theme updates (app blocks + metafields) | Liquid, theme app extensions |
| shopify-scripts-to-functions | Migrating deprecated Shopify Scripts to Functions | JavaScript, Shopify Functions |
Focus areas: computer vision (YOLO, CNNs) · medical & tabular ML · genomics · NLP & retrieval (RAG, embeddings) · model serving & MLOps