I am a highly skilled and innovative Data Analyst and Machine Learning professional with a passion for transforming complex data into valuable business insights. I specialize in building end-to-end data solutions that foster strategic decision-making, enhance profitability, and drive process automation.
My core strength lies in bridging the gap between deep technical processes and non-technical business audiences, ensuring that data-driven insights are not only discovered but are also clearly understood and acted upon.
- 🎓 I have an MSc in Data Analytics from the University of Portsmouth
- 💼 In my previous role, I developed a machine learning model that increased the 'hit rate' of electricity theft investigations by 29%.
- 📈 I also built a predictive failure model that led to a 15% improvement in grid reliability.
My toolbox includes a wide range of technologies for every stage of the data lifecycle, from analysis and modeling to visualization and deployment.
| Category | Skills |
|---|---|
| Programming & Databases | Python SQL R NoSQL (MongoDB) |
| Machine Learning | Scikit-learn TensorFlow Classification Regression Clustering Deep Learning (Transformers), NLT |
| Model Explainability (XAI) | LIME SHAP Auditing Model Logic |
| Data Viz & BI Tools | Power BI Tableau Microsoft Excel |
| Big Data & DevOps | Hadoop Spark Git GitHub |
Here are a few projects that showcase my ability to handle complex data challenges and deliver impactful results.
| Project | Description | Key Achievement |
|---|---|---|
| Feature Engineering Framework For Traffic Accident Prediction using XAI | Developed a novel framework to predict traffic accident severity using XGBoost, LGBM, and Random Forest. The core focus was translating complex model predictions into actionable road safety insights. | Achieved 92-93% accuracy and demonstrated how XAI can make sophisticated models transparent and interpretable for public safety initiatives. |
| Critical Audit of a Fake News Detector | This project went beyond simple detection. I used LIME not just to explain predictions, but to critically audit the model's logic. | Discovered the model was learning spurious correlations, identifying words like "Thursday" as predictors for "fake news," highlighting the critical need for XAI in building trustworthy AI. |
| Credit Card Fraud Detection with Big Data | Built a high-performance classification model to detect fraudulent transactions in a large dataset of 1.8 million records using TensorFlow, LGBM, XGBoost, and Transformers. | Engineered a robust solution that achieved an outstanding 97% to 100% accuracy across various algorithms, showcasing my ability to handle large-scale, imbalanced datasets. |