Chatbots that answer from your own documents, machine learning models you can trust, and the data pipelines that feed them. You get clear results, clean code and a short guide.
Job Success on Upwork
on every rated client job
place nationally at DEPI, as team lead
chatbot test questions right
I'm an AI engineer and data scientist from Egypt. I study Computer Science at Assiut University and work with clients on Upwork, with 100% Job Success and a 5.0 rating on every rated job.
I led my team to 3rd place nationally in the DEPI Data Science track with a satellite image model at 97% accuracy. What I care about most is work you can trust: every chatbot is checked with real questions, every model is tested without data leakage, and you get the results in plain English.
Three kinds of work, all done the same way: a clear plan, real tests, and results explained in plain English.
Chatbots that answer from your PDFs, manuals, policies or knowledge base, show the file and page for every answer, and say "not in the documents" instead of making things up.
Prediction, classification and forecasting models, plus data analysis with clear charts, reports and Power BI or Streamlit dashboards.
Data from websites, APIs, PDFs, invoices and scanned documents turned into clean Excel, CSV or Google Sheets, plus the script so you can run it again.
Native Arabic and fluent English, so I can also build chatbots and models for Arabic text and documents.
Every number below comes from the project's own evaluation, and the code is public on GitHub.
A RAG chatbot that answers from your own PDFs, shows the file and page for every answer, and says so when the answer is not there. Runs 100% on your own computer.
Invoices and receipts (PDFs, scans and phone photos) turned into clean, checked Excel data by a local AI model. No cost per page, and documents never leave the computer.
A deep learning model that recognises 10 types of land (crops, forest, rivers, roads, homes and more) in Sentinel-2 satellite images. I led the team to 3rd place in the national DEPI Data Science track.
Weekly demand forecasts with a low-high range for the 200 best-selling products of an online store, and a stock plan that covers the next 4 weeks 9 times out of 10.
An interactive dashboard on 367,189 real New York City property sales: market trends, neighbourhoods, a home price estimator with comparable sales, and rental yields by zip code.
1,067,371 raw order lines cleaned and modelled in SQL, a 4-page Power BI dashboard, and a 12-slide insights deck with 5 clear recommendations.
Send me your documents or data on Upwork and I'll reply with a plan.
For a repeat client on Upwork, three AI projects, all rated 5.0: a deep neural network built from scratch in NumPy (AUC 0.95), a reinforcement learning agent that learned to land a spacecraft (DQN vs a genetic algorithm), and machine learning notebooks in R.
Research: a model that predicts acute kidney injury in ICU patients (AUROC 0.965).
He delivers high-quality work and it is a pleasure to work with him.
Simple and clear, so you always know what is happening.
I ask questions and confirm the plan, timeline and price before I start.
On bigger jobs you get a small working sample early, so there are no surprises.
Tested on real data, and you get the numbers, not just "it works".
Revisions until it matches your brief. All the code is yours, with a short guide.
Send me what you have on Upwork and I'll reply with a plan, timeline and price, usually within a few hours.
