# Amr Mohammed — Machine Learning Engineer

> Machine Learning Engineer in Cairo, Egypt (open to remote), specializing in MLOps and cloud-native AI. I take models from notebooks to production — versioned data, tracked experiments, automated tests, repeatable deploys. The through-line of my work: measure first, and claim only what's measured.

This portfolio is a dependency-free static site. The case studies below are available as markdown — same content and same numbers as the HTML pages.

## Case studies

- [DocuMind — evaluation-first RAG](https://www.amr-mohammed.com/documind.md): a RAG platform with a 100-question hand-validated golden dataset, a three-metric eval harness, and a GitHub Actions gate that fails any PR that regresses retrieval. Single-hop recall@10 0.97; multi-hop 0.36 → 0.57 after a FlashRank + RRF reranker; 15/15 unanswerable questions refused; 4.82/5 mean answer faithfulness. Live demo: https://documind.amr-mohammed.com — HTML version: https://www.amr-mohammed.com/documind
- [OSS Contribution Copilot — measure before you build](https://www.amr-mohammed.com/oss-copilot.md): a supervised multi-agent system (in progress) that finds claimable open-source issues and drafts a human-approved claim comment. Built measurement-first: a 2,568-issue label audit broke four assumptions before any agent code — 90.7% of first type labels are author-applied, not maintainer judgment, and a 39.3-point authorship confound was caught and fixed in my own calibration set. The shipped artifact is a published MCP server (PyPI: oss-issues-mcp); the agent pipeline is in active development. HTML version: https://www.amr-mohammed.com/oss-copilot

## Other projects

- [la7za (لحظة)](https://github.com/AmrMohamed17/la7za): Arabic-first platform for personalized, interactive digital gift pages — built end-to-end solo, live with real payments. Next.js 16, TypeScript, Supabase, Paddle, fully RTL. Live: https://la7za.vercel.app
- [Local Product Finder](https://github.com/AmrMohamed17/Local-Product-Finder): graduation project, graded A+ — real-time product recognition (MobileNet) with a hybrid recommender (Sentence-BERT + LightFM), integrated into a mobile app with a cross-functional team.
- [Elevvo internship portfolio](https://github.com/AmrMohamed17/Elevvo_Internship): 6 end-to-end ML tasks — regression, clustering, multi-class classification, imbalanced-data loan approval (SMOTE), recommender systems (SVD), CNN traffic-sign recognition with transfer learning.

## Experience

- Research Assistant (contract, remote), freelance project affiliated with King Saud University — Dec 2025 to Apr 2026. Systematic searches across scientific databases with documented strategies; structured study summaries and a comparative evidence matrix.
- Machine Learning Engineer Intern, Elevvo Pathways — Jul to Aug 2025. Built and tuned 6+ ML models; designed reproducible pipelines with MLflow experiment tracking.

## Education and certifications

- BSc Computer Science, Ain Shams University (2021–2025); graduation project graded A+.
- Verified certifications: Advanced Learning Algorithms and Supervised ML: Regression & Classification (DeepLearning.AI · Stanford Online, 2024), Mathematics for Machine Learning (Imperial College London, 2024), CS50's Introduction to AI with Python (Harvard, 2024), CS50x (HarvardX, 2022).

## Contact

- Email: amrm88289@gmail.com
- GitHub: https://github.com/AmrMohamed17
- LinkedIn: https://www.linkedin.com/in/amr-mohammed01
- Resume (PDF): https://www.amr-mohammed.com/Amr_Mohammed_ML_Engineer_Resume.pdf