Komal Mustafa

Data Analyst

FSE • ZZP

Oct 2023 - Dec 2025 • 2 yrs 2 mos

I have been working in data analytics and machine learning for the past 4 months and honestly it has been one of the most intense and rewarding experiences of my life. I did not just follow tutorials and call it a day. I picked a real problem, built a real system, and saw it through from the very first line of code to a working trained model. The main project I am most proud of is a complete Flash Flood Early Warning System for Jeddah, Saudi Arabia. Jeddah has a serious flooding problem and I wanted to build something that could actually help. I collected over 350,000 rows of real hourly climate data going all the way back to 1985 using open weather APIs. I then cleaned the data, handled missing values, removed errors, and engineered over 35 custom features including rolling rainfall sums, soil moisture indexes, atmospheric instability scores, and seasonal risk flags. After that I built and trained a CNN-LSTM deep learning model in PyTorch that combines convolutional layers for pattern detection with LSTM layers for time sequence memory. The model hit 99.96 percent accuracy on the first epoch which honestly surprised even me. Every single step of this project I did myself inside VS Code with a custom Python virtual environment. I wrote separate scripts for data collection, labeling, feature engineering, model architecture, and training. I treated it like a production codebase, not a notebook experiment.Outside of that project I have been consistently building my skills across the full data stack. I work comfortably with Python, Pandas, NumPy, Scikit-Learn, PyTorch, Matplotlib, Seaborn, Plotly, SQL, Tableau, Power BI, and Google Sheets. I am currently finishing the Streamlit dashboard for the flood system which will display live risk alerts on an interactive map.I am early in my career but I work like someone who has been doing this for years. If you give me a problem I will find a way to solve it properly.