MLOps & ML Systems Engineer | Model Serving · Docker · CI/CD · AWS · Monitoring
I design and operate end-to-end ML systems — from reproducible training pipelines to containerized inference, automated CI/CD, cloud deployment, and production monitoring. My background in applied physics and mathematics informs how I approach modeling, while my engineering practice is grounded in MLOps: version control for data and models, infrastructure as code, and observable, maintainable services.
End-to-end MLOps platform for spell correction. DVC-managed data and model versioning, Docker containerization, GitHub Actions CI/CD with GHCR publishing, and Flask-based model serving. Demonstrates a full ML lifecycle from data ingestion through validated deployment.
Reusable production inference scaffold built on BentoML. Includes Docker packaging, CI/CD pipeline, health checks, structured logging, and monitoring hooks. Designed as a starting point for deploying ML models behind a production-grade API.
Operational data and model drift detection. Computes PSI, KS statistic, Wasserstein distance, and Jensen-Shannon divergence per feature. Exposes Prometheus-compatible metrics, includes alerting logic, and ships with synthetic drift tests.
- Production-grade ML inference templates
- Drift and model-quality monitoring
- Infrastructure-as-code for ML services
- Physics-informed ML systems
GitHub Statistics
Site: iblahlou.github.io LinkedIn: Ibrahim Lahlou GitHub: IbLahlou