amr@production:~$ ✓ open to work

deploy — career.sh

MLOPS · CLOUD-NATIVE AI · RAG

I'm Amr Mohammed. I take models from notebooks to production — versioned data, tracked experiments, automated tests, repeatable deploys.

View projects Resume ↓ GitHub ↗

▾ scroll to deploy

[01] ./projects

Things I've built

The model is the easy 20%. These projects are about the other 80% — the engineering that makes AI dependable.

● IN PROGRESS

OSS Contribution Copilot

A supervised multi-agent system that finds claimable issues in OSS repos and drafts the claim comment — human-approved, never autonomous. Built measurement-first: a 2,568-issue label audit broke four assumptions before any agent code (90.7% of type labels are author-applied, not maintainer judgment). The MCP server is published on PyPI; the agent pipeline is in progress.

GitHub REST API · 2,568-issue label audit · needs-info calibration across 5 ecosystems · LLM readiness judge · MLflow · MCP server on PyPI (oss-issues-mcp) · LangGraph + Postgres checkpointer (in progress)
2,568issues audited before any agent code
90.7%of type labels author-applied
39.3 ptsauthorship confound caught · fixed
4 toolsMCP server · PyPI + MCP Registry
read the case study → pypi ↗ source ↗
● LIVE

DocuMind — Evaluation-First RAG

RAG treated as a measured system: a 100-question hand-validated golden dataset, a three-metric eval harness, and a CI gate that fails any PR that regresses retrieval. Multi-hop lagged — I traced every miss to ranking and fixed it with a FlashRank reranker + rank fusion: 0.36 → 0.57 recall@10, single-hop held. Deployed and live on AWS.

FastAPI · Gemini embeddings (768d) · PostgreSQL/pgvector · FlashRank reranker · golden-set eval harness · GitHub Actions CI gate · Docker
0.90 / 0.97single-hop recall @3 / @10 · held
0.36 → 0.57multi-hop recall @10 · reranked
15/15unanswerable refused
4.82/5answer faithfulness · LLM-judged
live demo ↗ read the case study → source code ↗
● LIVE

la7za (لحظة)

Arabic-first platform for personalized, interactive digital gift pages — built end-to-end solo and shipped with real payments. Next.js 16 + TypeScript, Supabase, Cloudinary, Paddle, fully RTL.

nextjstypescriptsupabasepaddle
live product ↗ source ↗
GRADUATION PROJECT · A+

Local Product Finder

Real-time product recognition (MobileNet, 99% accuracy on the project's test set) with a hybrid recommender — Sentence-BERT semantic similarity + LightFM collaborative filtering — integrated into a mobile app with a cross-functional team.

mobilenetsentence-bertlightfmtensorflow
source ↗
INTERNSHIP WORK

6 End-to-End ML Tasks

Complete internship portfolio at Elevvo: regression, clustering, multi-class classification, imbalanced-data loan approval (SMOTE), recommender systems (SVD), and CNN traffic-sign recognition with transfer learning.

scikit-learnxgboosttensorflowsurprise
source ↗
[02] ./experience

Where I've worked

Dec 2025 — Apr 2026

Research Assistant (Contract, Remote)

Freelance — project affiliated with King Saud University
  • Systematic searches across scientific databases with documented strategies and inclusion/exclusion criteria
  • Structured study summaries and a comparative evidence matrix consolidating findings and metrics
Jul 2025 — Aug 2025

Machine Learning Engineer Intern

Elevvo Pathways
  • Built and tuned 6+ ML models (regression, classification, CNNs), applying SMOTE for class imbalance and hyperparameter optimization
  • Designed reproducible ML pipelines for preprocessing and feature engineering, integrating MLflow for experiment tracking and versioning
  • Applied dimensionality reduction (SVD) and transfer learning to improve model performance
[03] ./education

Academic background

A strong academic track that bridges a computer science degree with a high-school record built on strong scores.

Oct 2021 — Jul 2025

BSc Computer Science

Ain Shams University
  • Graduation project (Local Product Finder) graded A+
  • Built a practical foundation in software, data, and machine learning
Oct 2019 — Jul 2021

High School Diploma

Tiba American College
  • Completed the American high school curriculum with a 3.95 GPA
  • SAT: 1270 · ACT: 32
  • Strengthened the academic base that led into computer science and machine learning
[04] ./certifications

Verified credentials

Every certificate links to its issuer's verification page.

Advanced Learning Algorithms
DeepLearning.AI · Stanford Online — 2024
verify ↗
Supervised ML: Regression & Classification
DeepLearning.AI · Stanford Online — 2024
verify ↗
Mathematics for Machine Learning
Imperial College London — 2024
verify ↗
CS50's Introduction to AI with Python
Harvard University — 2024
verify ↗
CS50x: Introduction to Computer Science
HarvardX — 2022
verify ↗
Machine Learning Workshop '25
FCIS, Ain Shams University — 2025
[05] ./skills

What I work with

MLOPS & DEVOPS

  • Docker
  • GitHub Actions (CI/CD)
  • DVC
  • MLflow
  • AWS — EC2, S3, RDS
  • Linux

AI & GENAI

  • Multi-Agent Systems
  • LangGraph
  • LangChain
  • MCP (Model Context Protocol)
  • RAG pipelines
  • Vector DBs (pgvector, ChromaDB)
  • Computer Vision
  • NLP

BACKEND & DATA

  • Python
  • FastAPI
  • SQL
  • REST APIs
  • Pytest
  • Async APIs

CORE ML

  • TensorFlow
  • PyTorch
  • Scikit-learn
  • Pandas / NumPy
[06] ./contact

Let's build something that ships.

Open to Machine Learning Engineer and AI Engineer roles — remote or Cairo. If your team cares about AI that survives contact with production, we'll get along.

amrm88289@gmail.com Resume ↓ LinkedIn ↗ GitHub ↗