I am a Master's student in Berlin bridging the gap between raw data and intelligent decision-making. I specialize in building robust NLP pipelines, predictive models, and geospatial analyses that solve real-world problems across business, healthcare, and urban planning.

• Currently writing my Master's thesis on Cross-Lingual Transfer in Multilingual LLMs (BERT, mBERT, XLM-RoBERTa)
• Recently published a satellite-based Urban Canopy & Heat Analysis of Riyadh using Google Earth Engine
• Authored research on Mental Health Detection using BERT & RoBERTa
• Open to Working Student & Entry-Level roles in ML/AI, Data Science, and Geospatial Analytics

Satellite-derived priority map for urban greening interventions in Riyadh.

• Tech: Google Earth Engine, Sentinel-2, Landsat 8/9, ESA WorldCover, Python.
• Impact: Quantified +27% vegetation cover growth (2019–2025) and identified 246 km² of high-priority planting zones — a 2.5× expansion envelope on current canopy.
• Highlight: Multi-sensor fusion (optical NDVI + thermal LST + land-cover masking) producing planning-ready intervention zones.

End-to-end climate analysis mapping how El Niño reorganises rainfall worldwide, with a validated regional prediction model.

• Tech: xarray, Cartopy, Scikit-learn, NOAA ERSSTv5 & GPCP, Python.
• Impact: Reproduced every canonical El Niño teleconnection from raw data; a model trained excluding the 2015/16 super-El-Niño still anticipated its rainfall pattern (DJF correlation up to 0.91).
• Highlight: Diagnosed that the model captures rainfall timing but under-predicts extreme magnitude — and that ENSO skill decays from the Pacific core outward, weakest where the Indian Ocean Dipole dominates.

Authored research benchmarking Transformer models against Deep Learning baselines.

• Tech: BERT, RoBERTa, LSTM, PyTorch.
• Impact: Achieved >99% accuracy in detecting depressive language on social media.
• Highlight: Addressed ethical AI challenges and data bias.

End-to-end classification pipeline identifying key service differentiators.

• Tech: Random Forest, GridSearch, Feature Importance Analysis.
• Impact: Achieved 96% Test Accuracy and identified In-flight Wi-Fi as a top ROI driver.

Predictive system to support data-driven retention strategies.

• Tech: XGBoost, SMOTE (Class Imbalance), Scikit-learn.
• Impact: Optimized model for Recall to capture high-risk customers for intervention.