π Graduating in Actuarial Sciences β Hacettepe University
π Passed Level 2 Actuarial Exam β Turkish Actuarial Society
πΌ Assistant Actuary β Hands-on experience in insurance portfolio analytics & pricing
π¬ TΓBΔ°TAK 2209-A Research Grantee β Funded undergraduate research in quantitative finance
I combine actuarial rigor with modern data science to build reproducible, end-to-end analytical pipelines β from risk modeling and predictive analytics to interactive decision-support dashboards.
Currently seeking opportunities in actuarial consulting, insurance analytics, and quantitative risk management.
Languages & Querying
Machine Learning & AI
Supervised & Unsupervised Learning Β· Deep Learning (LSTM Β· GRU Β· CNN Β· Transformer) Β· Time Series Forecasting Β· Classification & Regression
Actuarial & Statistical Modeling
GLM (Poisson Β· Gamma Β· Logistic) Β· Life & Non-Life Insurance Modeling Β· Loss Reserving Β· Survival Analysis Β· Frequency-Severity Models Β· Risk Classification
Visualization & Reporting
| Project | Domain | Highlights |
|---|---|---|
| π¬ MC-AWARE | Deep Learning Β· Finance | TΓBΔ°TAK 2209-A grant Β· 26 experiments Β· 6 DL architectures Β· 378+ configs Β· Discovered anti-predictive behavior in BIST (p β 10β»ΒΉβ΄) |
| π Direction Forecasting | Academic Research | ARIMA vs LSTM vs 1D-CNN on BIST & pension funds β majority class illusion in Turkish markets |
| π ADAS Pricing Paradox | Actuarial Pricing | 100K synthetic policies Β· Poisson + Gamma GLM Β· Proves ADAS safety β lower premiums |
| π ADAS Vol.2 | Advanced Actuarial | 200K policies Β· Gini Index Β· Lift Charts Β· Interaction terms Β· Power BI dashboard |
| π Risk Scoring Dashboard | Interactive Analytics | R Shiny + Logistic GLM Β· AUC 0.828 Β· Real-time claim probability scoring |
π¬ Anti-Predictive Discovery β Proved that 118/120 deep learning configurations systematically predict the opposite market direction on BIST, driven by macro-variable concept drift (USDTRY, Oil, TCMB rate shifts).
π End-to-End Actuarial Pipeline β Designed a full pricing workflow: synthetic data generation β feature engineering (SQL) β GLM modeling (R) β interactive Power BI dashboards β uncovering the ADAS Pricing Paradox in Turkish auto insurance.
π― Real-Time Decision Support β Built an interactive Shiny risk scoring platform achieving AUC 0.828, bridging actuarial modeling with operational decision-making.
π 738+ Trained Models β Across all projects: 513 pension fund models + 225 BIST models, each with multi-seed cross-validation for full reproducibility.
π§ AI & Advanced Analytics β Explainable AI (XAI) for insurance decisions Β· NLP for claims processing Β· Gradient Boosting (XGBoost, LightGBM) for risk scoring
π Actuarial Innovation β Telematics-based pricing Β· Climate risk modeling Β· Stochastic reserving Β· IFRS 17 reporting automation
π§ Data Engineering β ETL pipeline design Β· Cloud-based actuarial workflows Β· Automated reporting systems
π Quantitative Finance β Portfolio optimization Β· Credit risk modeling Β· Algorithmic trading strategies
π Professional Growth β Advanced actuarial exams Β· SOA/IFoA pathway exploration Β· Kaggle competitions
"It is better to be on hand with ten men than absent with ten thousand."
β Timur (Tamerlane)
Open to actuarial analyst, insurance analytics, and quantitative risk roles β let's connect!