Mission249
A gamified volunteering platform that helps Sudanese youth rebuild their communities through small missions and teamwork. It took 3rd place at the Code for Sudan Hackathon.
View on GitHub →§1 About
Asmaa Farouq
Computer Science & AI student · Cairo University
I'm a third-year Computer Science and AI student at Cairo University. I like building practical things — web apps, ML pipelines. Outside of code, I'm curious about science, history, and the humanities.
§2 Now
This month
- Learning: computer vision
- Reading: Hands-On Machine Learning by Aurélien Géron
- Practicing: Pytorch and Linux
§3 Software
Maly
A personal finance app in Django, structured around the Observer, Facade, and Strategy design patterns and documented with Sphinx and pdoc.
View on GitHub →
GeziraPal
A platform where farming communities can share and find practical knowledge on crops and sustainable agriculture.
View on GitHub →
Recipe Finder
Started as plain HTML/CSS/JS, then we rebuilt it in Django with AJAX and a SQLite database, split into multiple apps.
View on GitHub →§4 Machine learning
Most of what I know about ML I learned by entering competitions and getting things wrong in notebooks. Everything below lives in my Applied Machine Learning Lab repo.
Recognitions
INEGI GCIM Vegetation Mapping Challenge
9th of 69 teams · SilverPredicting the vegetation class of regions in Mexico from spectral, radar, and terrain features (45 classes). We compared several boosted-tree models and averaged LightGBM and XGBoost for the final submission.
WiDS Datathon 2025
Predicting ADHD and sex in children from about 1,200 samples with over 19,000 features. The hard part was class imbalance, so we used stratified K-fold with downsampling and tuned decision thresholds for F1.
Zindi GeoAI Aquaculture Pond Identification
Classifying aquaculture ponds from satellite imagery: EDA, a LightGBM model, SHAP to understand which features mattered, and Kedro to keep the pipeline organised.
Projects
Kepler Exoplanet Classification
Classifying NASA Kepler objects as confirmed planets, candidates, or false positives. After EDA and comparing a few models, LightGBM reached about 92% accuracy on the test set.
Airbnb New User Booking Prediction
Predicting where a new Airbnb user will make their first booking, using their profile data. I did EDA and feature engineering, then compared LightGBM, XGBoost, and CatBoost, evaluating with NDCG.
§5 Teaching
Technical Member, Google DSC – Khartoum University
- I tutor students in AI concepts and lead hands-on sessions, breaking technical topics down into clear explanations.
- I help run the program and its media output using Google Workspace, working with volunteers spread across different places.
- I explain our technical activities to a wide student audience, both in writing and meetings.
AI Instructor, DSC – Cairo University
- I ran hands-on AI/ML workshops with assignments and mini-projects, which got more students actively taking part.
- I wrote the lesson plans and tried to explain difficult ideas in a way that makes sense to beginners.
§6 Toolbox
- Languages
- Python, C++, JavaScript, SQL
- Web
- Django, HTML/CSS, AJAX
- ML
- pandas, NumPy, Matplotlib, Seaborn, scikit-learn
- Tools
- Git/GitHub, Linux
§7 Outside the code
When I'm not coding, I'm usually reading about science or history. I like understanding how people figured things out in the first place, and how ideas moved between places and centuries. The humanities keep reminding me that the things we build are always for people.
I also crochet. It's nice to make something with my hands that I can actually hold, and following a pattern one stitch at a time isn't that different from debugging.