1.90% DER
Arabic text diacritization with a 9-model ensemble and BiLSTM architecture.
ML/NLP + Software Engineering
I build ML/NLP systems and production-grade software with a QA-first mindset. My recent work spans Arabic diacritization, churn intelligence at scale, and credit risk modeling.
From competition-grade sequence models to production data pipelines, I ship systems that are measurable, explainable, and resilient.
Recent outcomes
Telecom churn pipeline in PySpark.
Credit lending model vs approve-all baseline.
Regression failures resolved at Siemens DISW.
Highlights
Evidence-driven outcomes across research, data engineering, and quality automation.
Arabic text diacritization with a 9-model ensemble and BiLSTM architecture.
Portfolio value lift from custom threshold tuning on credit lending models.
TCL command coverage increase after migrating 200+ test cases.
Issues reported at Siemens DISW, including 4 critical findings.
What I Do
I balance model design, data systems, and rigorous QA to ship reliable ML.
BiLSTM ensembles, frozen transformers, and ablation-driven experiments.
AraBERT, Word2Vec, PyTorchDistributed pipelines, custom MapReduce algorithms, and model lifecycle tracking.
PySpark, MLflow, FastAPIAutomation libraries, regression analysis, and cross-team release readiness.
Squish, Cypress, EDA toolsSelected Projects
Experience
Hands-on delivery across Siemens DISW and academic projects.
Skills
Education
Oct 2021 - Jul 2026 (Expected)
Grade: 88.82%
Vice-Head, Software and Automation Development
Cairo University Racing Team (CURT), Sep 2023 - Present
Contact
Reach out for ML/NLP projects, data systems work, or quality engineering collaboration.