Daniele Zago

Daniele Zago

Algorithm Engineer

Algorithm Engineer @Timefold, Ghent

Research-oriented data and decision scientist bridging statistics, machine learning, and mathematical optimization. I design and implement end-to-end solutions spanning stochastic optimization algorithms, ML models, and production software on enterprise stacks, with a track record in both academic research and industrial deployment.
Contact

Education

Academic training in statistics, optimization, and scientific computing.

  1. BSc 2016 — 2019

    Bachelor of Science (B.Sc.)

    Statistics for Technology and Sciences

    University of Padova Padova, Italy

    • Final grade: 110/110 cum Laude
    • GPA: 29.2/30
    • Thesis topic: Applied Bayesian modelling
  2. MSc 2019 — 2021

    Master of Science (M.Sc.)

    Statistical Sciences

    University of Padova Padova, Italy

    • Final grade: 110/110 cum Laude
    • GPA: 29.5/30
    • Thesis topic: Bayesian nonparametric models

    During MSc

    Summer school July 2020

    Summer School in Mathematics

    University of Perugia

    Functional analysis and mathematical statistics

  3. PhD 2021 — 2024

    Doctor of Philosophy (Ph.D.)

    Statistical Sciences

    University of Padova Padova, Italy

    • Advisors: Prof. Giovanna Capizzi (University of Padua), Prof. Peihua Qiu (University of Florida)
    • Research topics: Online outlier detection, stochastic optimization

    During PhD

    Summer school October 2022

    13th International School on Efficient Scientific Computing

    INFN, Bertinoro

    Efficient C++ and GPU programming with CUDA

Experience

Experienced in mathematical optimization, machine learning, and statistical computing, with applications in operations research and industrial statistics.

Algorithm Engineer

Timefold

Ghent, Belgium

2026 — Present

Algorithm engineering at Timefold, working on Java-based optimization software for planning and decision-making problems.

Technologies: Java

Data Scientist

OPTIT S.r.l

Bologna, Italy

2024 — 2026
  • Research and development of solutions to the periodic vehicle routing problem with clustered routes for multi-day strategic fleet management, including model deployment and development for interactive visualizations (Python, Java, Hexaly, Gradle, Docker, TypeScript, Next.js)
  • Improved minimum cost flow approaches for crossdock facility selection when serving customers from warehouses with potential crossdock drop-off of goods (Hexaly, Xpress Mosel, Docker)
  • Developed forecasting and anomaly detection pipelines for power plant energy production/consumption and heat and electricity generation using machine learning methods (Python, scikit-learn, MLFlow)

Ph.D. Student in Statistical Sciences

University of Padua

Padova, Italy

2021 — 2024

During my Ph.D. in Statistical Sciences at the University of Padua (supervisors: Prof. Giovanna Capizzi and Prof. Peihua Qiu), I focused on advancing online outlier detection and stochastic optimization methods for statistical process monitoring.

My research resulted in publications in top-tier journals including the Journal of Quality Technology, Statistics & Computing, Quality Engineering, and the Journal of Statistical Software. I also developed the StatisticalProcessMonitoring.jl package, a comprehensive Julia framework for real-time process monitoring.

Additionally, I applied my expertise in real-world contexts through consulting projects on sequential monitoring systems for structural health monitoring, involving algorithm development in R and C++.

Thesis: Advanced Statistical Process Monitoring using Simulation-Based Algorithms (2024)

Research stay: University of Florida (Jan 2023 - Dec 2023), supervised by Prof. Peihua Qiu

Statistical Consultant

Expin S.r.l.

Padova, Italy

2023

Developed a sequential monitoring system in R and C++ for anomaly detection in vibrating string strain gauges, addressing critical challenges in structural health monitoring of civil infrastructure.

Designed a novel algorithm for optimal alarm threshold selection using stochastic optimization techniques, enabling adaptive and efficient detection of structural anomalies while minimizing false alarms.

The project combined statistical process control, computational efficiency (C++ implementation), and modern optimization theory to deliver a production-ready monitoring solution.

Teaching Assistant

University of Padua - Dept. of Developmental Psychology

Padova, Italy

2022

Delivered lectures on factor models and questionnaire analysis for graduate students in psychology, bridging statistical theory with practical applications in psychological research.

Conducted hands-on R programming laboratory sessions, teaching students statistical computing skills for data analysis and visualization in psychological studies.

Supported students in developing their research projects, providing guidance on statistical methodology and computational implementation.

Academic Tutor

University of Padua - Dept. of Statistical Sciences

Padova, Italy

2017 — 2019

Provided calculus lessons and interactive workshops for undergraduate students in the Statistics program, helping students develop strong mathematical foundations for their statistical education.

Conducted regular tutoring sessions covering topics in mathematical analysis, including limits, derivatives, integrals, and series, with emphasis on problem-solving techniques.

Organized guided exercise sessions to prepare students for final examinations, improving pass rates and student confidence in mathematical reasoning.

Publications & Software

Research in statistical computing, mathematical optimization, and decision sciences.

Software

StatisticalProcessMonitoring.jl

StatisticalProcessMonitoring.jl Julia

Journal of Statistical Software

A Julia package for real-time statistical process monitoring, integrating advanced algorithms and control charts to handle complex data types, such as sequential data, functional data, and structured observations.

Click to read more

StatisticalProcessMonitoring.jl

StatisticalProcessMonitoring.jl Julia

Journal of Statistical Software

A Julia package for real-time statistical process monitoring, integrating advanced algorithms and control charts to handle complex data types, such as sequential data, functional data, and structured observations.

Paper Source Code

Published Articles

A doubly-stochastic constrained optimization algorithm

A doubly-stochastic constrained optimization algorithm

Journal of Quality Technology

I designed a novel stochastic optimization algorithm with stochastic constraints, aimed at optimizing control chart tuning parameters with greater efficiency compared to traditional numerical methods.

Click to read more

A doubly-stochastic constrained optimization algorithm

A doubly-stochastic constrained optimization algorithm

Journal of Quality Technology

I designed a novel stochastic optimization algorithm with stochastic constraints, aimed at optimizing control chart tuning parameters with greater efficiency compared to traditional numerical methods.

View Article
An improved bisection-type algorithm for control chart calibration

An improved bisection-type algorithm for control chart calibration

Statistics and Computing

I developed a modified bisection algorithm for computing control limits in complex settings where standard methods are inefficient. The approach removes the need for a predefined search range and scales efficiently to multi-chart scenarios.

Click to read more

An improved bisection-type algorithm for control chart calibration

An improved bisection-type algorithm for control chart calibration

Statistics and Computing

I developed a modified bisection algorithm for computing control limits in complex settings where standard methods are inefficient. The approach removes the need for a predefined search range and scales efficiently to multi-chart scenarios.

View Article
Alternative parameter learning schemes for monitoring process stability

Alternative parameter learning schemes for monitoring process stability

Quality Engineering

I formalized the theoretical behavior of parameter learning schemes in relation to outlier detection performance in control charts. This work enabled the generalization of alternative parameter learning methods, leading to a more efficient and accurate detection scheme compared to traditional approaches.

Click to read more

Alternative parameter learning schemes for monitoring process stability

Alternative parameter learning schemes for monitoring process stability

Quality Engineering

I formalized the theoretical behavior of parameter learning schemes in relation to outlier detection performance in control charts. This work enabled the generalization of alternative parameter learning methods, leading to a more efficient and accurate detection scheme compared to traditional approaches.

View Article
A general framework for monitoring mixed data

A general framework for monitoring mixed data

Journal of Quality Technology

We developed a general methodology to monitor processes involving mixed-type data (continuous, ordinal, categorical), common in real-world applications. The method enables effective sequential monitoring under serial correlation.

Click to read more

A general framework for monitoring mixed data

A general framework for monitoring mixed data

Journal of Quality Technology

We developed a general methodology to monitor processes involving mixed-type data (continuous, ordinal, categorical), common in real-world applications. The method enables effective sequential monitoring under serial correlation.

View Article

Under Review & In Preparation

Monitoring of complex geometrical shapes

Monitoring of complex geometrical shapes Submitted

I developed an innovative statistical quality control method for detecting shape defects in complex geometries obtained via additive manufacturing. I introduced a novel nonparametric control chart based on kurtosis analysis which provides superior defect detection for 3D-printed objects compared to existing approaches.

Click to read more

Monitoring of complex geometrical shapes

Monitoring of complex geometrical shapes Submitted

I developed an innovative statistical quality control method for detecting shape defects in complex geometries obtained via additive manufacturing. I introduced a novel nonparametric control chart based on kurtosis analysis which provides superior defect detection for 3D-printed objects compared to existing approaches.
Likelihood-ratio monitoring of processes with mixed data

Likelihood-ratio monitoring of processes with mixed data In Prep

Journal of Quality Technology

We developed a likelihood-ratio methodology to monitor processes involving continuous and categorical data. The approach makes use of adaptive kernel density estimation in order to approximate the likelihood ratio, thus providing efficient detection power.

Click to read more

Likelihood-ratio monitoring of processes with mixed data

Likelihood-ratio monitoring of processes with mixed data In Prep

Journal of Quality Technology

We developed a likelihood-ratio methodology to monitor processes involving continuous and categorical data. The approach makes use of adaptive kernel density estimation in order to approximate the likelihood ratio, thus providing efficient detection power.

Conference Presentations

Selected presentations on stochastic optimization, statistical process monitoring, and decision sciences.

Nov 2025
Invited Seminar

Efficient algorithms for the calibration of control limits

Università degli Studi di Padova , Padova, Italy

Sep 2025
Invited Talk

Optimal constrained design of control charts using stochastic approximations

ENBIS-25 Conference , Piraeus, Greece

Oct 2023
Invited Talk

Optimal constrained design of control charts using stochastic approximations

2023 INFORMS Annual Meeting , Phoenix, AZ, USA

Sep 2022
Poster

Profile monitoring based on adaptive parameter learning

Statistical methods and models for complex data , Padova, Italy

Jun 2022
Poster

Bayesian nonparametric multiscale mixture models via Hilbert-curve partitioning

2022 ISBA World Meeting , Montréal, Canada