I'm currently a Data Science Intern at BEAT (Better Environment and Transportation) since June 2025 🌍.
Actively seeking long-term Data Science/AI roles in 2025!
🎯 Aspiring Data Scientist & AI Engineer, certified by DataCamp
🎓 BSc in Astrophysics with Space Science (1st Class)
I started by studying galaxies, but found my real passion in data-driven discovery. Today, I design end-to-end ML projects from raw data pipelines to interactive dashboards that reveal insights and drive smarter decisions.
Languages
Python | SQL (PostgreSQL, Snowflake) | Bash | LaTeX
Data Science & ML
Pandas NumPy Matplotlib Seaborn SciPy
Scikit-learn XGBoost PyTorch SHAP LIME
emcee (Bayesian Inference)
Tools & Platforms
Git & GitHub | Docker | VS Code | Streamlit | MLflow
✨ Currently focused on:
- ⚡ Scientific Computing & Simulation : Transportation sector
- 🧠 LLMs & Generative AI: Hugging Face, fine-tuning transformers
- 📈 Time-Series Forecasting: LSTM, CNNs, SARIMAX for predictive analytics
- 🔎 Interpretable ML: SHAP, LIME for model transparency
📊 End-to-End Customer Churn Prediction
- Built an ML pipeline (Logistic Regression, RF, XGBoost, Voting Classifier) → ROC-AUC: 0.87
- Applied KMeans & HDBSCAN clustering with UMAP to segment customers into actionable personas
- Used SHAP explainability to reveal key churn drivers (tenure, contract type, fibre optic service)
- Deployed interactive Streamlit App for business users
- Containerised workflow with Docker, reducing setup time by >80% and ensuring reproducibility
- Tracked experiments, metrics, and models with MLflow, improving transparency and versioning across the pipeline
- Designed an A/B testing simulator with Chi-Square tests to measure the impact of retention strategies
- Big Data Platforms (PySpark, AWS, Databricks)
- Deep Reinforcement Learning
- 📬 Email: tsenghintsz@gmail.com
"The most exciting phrase to hear in science, the one that heralds new discoveries, is not 'Eureka!' but 'That's funny...'" — Isaac Asimov