Mathematician and Machine Learning Engineer working on probabilistic models, production-oriented ML systems and quantitative applications.

I hold a Master’s Degree in Mathematics and I combine cybersecurity research with applied AI and data science. My work is focused on reliable machine learning, probabilistic modeling and quantitative risk, with an emphasis on reproducible pipelines and deliverable solutions beyond notebooks.

[LinkedIn](URL: www.linkedin.com/in/federico-lancini-71604724b ) · [Email](mailto: federico.lancini@unicatt.it or federicolancini@alice.it)

Temporal football valuation using a semi-Markov formulation, survival analysis and StatsBomb 360 contextual data.

Survival Analysis · Probabilistic Modeling · Sports Analytics

See the repository for methodology, experiments and results.

End-to-end predictive maintenance pipeline with MLflow tracking, DVC data versioning, FastAPI service and containerization.

MLOps · Production ML · Imbalanced Classification

Open the repository for implementation details and deployment workflow.

Quantitative risk lab for portfolio modeling with GARCH, Monte Carlo scenarios, VaR/ES and optimization.

Quantitative Risk · Portfolio Optimization · Stress Testing

Refer to the repository for notebooks, validation and risk metrics.

Network traffic classification and DDoS detection research using PCAP feature extraction, early attack detection and domain-shift evaluation.

Cybersecurity ML · Imbalanced Classification · Domain Shift

See the repository for data, models and research findings.

Probabilistic Modeling · Machine Learning Engineering · MLOps · Survival Analysis · Quantitative Risk · Cybersecurity ML

Languages: Python, SQL Machine Learning: PyTorch, scikit-learn, XGBoost, CatBoost, pandas, NumPy MLOps and Backend: FastAPI, MLflow, DVC, Docker, GitHub Actions, pytest Quantitative Methods: Survival Analysis, Monte Carlo simulation, Portfolio Optimization, Value at Risk, Expected Shortfall, GARCH

This profile provides a high-level overview. Each repository contains its own documentation, methodology, experiments and reproducibility instructions.

Developing reliable machine learning systems, probabilistic models and production-oriented pipelines for quantitative and risk-aware applications.

Currently working on these projects:

• SQIL for LLM fine-tuning and distillation
• PINNs for solving experimental physics equations