Tomislav Suhina

Experience

Professional Experience

Freelance Decision Scientist (ZZP)

Independent Consultant | April 2023 to Present

Independent data science consultant (registered ZZP) serving diverse clients from startups to Fortune 500 companies. Specializing in production ML systems, MLOps, and end-to-end solution delivery across multiple industries.

Selected Client Engagements:

Heineken

Building production ML frameworks for IoT and logistics optimization: - End-to-end ML pipelines for equipment monitoring and predictive analytics - Logistics optimization solutions with scalable architecture - Production deployment with monitoring and maintenance frameworks

Rabobank

Delivered production ML systems for business lending analytics: - RAG Knowledge Retrieval System: Information retrieval solution supporting business decisions, deployed to production - Customer Retention Model: Predictive modeling with A/B testing framework, validated and deployed to production - MLOps Blueprint: Comprehensive deployment framework (CI/CD, drift detection, data quality, monitoring)

Lead Data Scientist Global Commerce & Finance

Heineken | August 2022 to March 2023

Led interdisciplinary team of 15+ direct reports (6 data scientists, 2 front end engineers, 2 data engineers, 3 MLOps engineers, and others) and served as interim product owner for Heineken's Global Data Harmonisation project.

Delivered Media Mix Modeling, one of Heineken's most successful analytics projects with significant captured business value (tens of millions in first year alone). Highly impactful and used to this day. Also delivered Raw Material Demand Forecasting and Cost Benefit Simulation.

Data Scientist

Heineken | October 2019 to July 2022

Individual contributor who learned end to end delivery by wearing all the hats, from model architecture and data engineering to project management and stakeholder management. Often served as technical lead. Built solutions that became highly impactful and are used to this day.

Key projects:

Financial KPI Forecasting & Time Series Best Practices: Full end to end ownership covering model architecture, development, MLOps, data engineering, project management, stakeholder management, and delivery scoping. Served as technical lead. Established time series best practices for the organization. Highly impactful, with the team later taking ownership and the system still in use to this day.

Equipment Spare Parts Allocation: Initiated and built with full end to end ownership across technical, project management, and stakeholder management aspects. Served as technical lead. The project became a driving force for operational improvements across the organization with massive opportunity. Highly impactful and still live and used to this day.

Also explored diverse applications including equipment failure detection (IoT), churn prediction, bottle design recommender, warehouse bin allocation, delivery route optimization, and data quality automation. Delivered workshops, trainings, and mentoring across the organization.

Learned by doing everything. Built projects that are highly impactful and used to this day. Later returned as contractor for various improvements on original projects.

Data Scientist

Xccelerated (Xebia) @Heineken | September 2018 to October 2019

Joined newly established analytics department in a fast paced, startup minded environment. Experimented with use cases to figure out what delivered value, facilitated business enablement through hands on analytics trainings, and created materials for analytics translators to support utilization of analytical solutions.

Fin/ID Tech Consultant

UL | July 2017 to September 2018

Contributed to Public Key Infrastructure design, documenting and testing for different governments. Handled pre launch integration testing for leading mobile payment service provider, test automation for bank cards, and technical writing.

Education

Dr. Sc. Physical Chemistry

University of Amsterdam | 2012 to 2017

Research focus: Fundamental understanding of molecular motion under confined conditions. Designed and synthesized molecules, attached them to surfaces, and observed their reaction to contact induced confinement. Used extensive modeling (MATLAB) and a wide variety of experimental methodologies to characterize molecular behavior.

Publications: Multiple papers in highly cited international peer reviewed journals

Skills gained: Scientific rigor, experimental design, extensive modeling, data analysis, technical writing, presenting complex concepts

MSc. Applied Organic Chemistry

University of Zagreb, Croatia | 2010 to 2012 | summa cum laude

BSc. Applied Chemistry

University of Zagreb, Croatia | 2007 to 2010 | cum laude

Technology Stack

Programming Languages

Python • SQL • Bash/Unix scripting

Machine Learning & AI Frameworks

  • Classical ML: Scikit-learn, XGBoost, CatBoost, SparkML
  • Deep Learning: PyTorch, PyTorch Lightning, fastai
  • LLM & RAG: OpenAI, HuggingFace, LangChain, LangGraph, LlamaIndex, ChromaDB, FAISS
  • Time Series: Prophet, PyMC, DARTS, Nixtla, Stumpy, Ruptures
  • Causal Inference: PyMC, CausalNex, NetworkX
  • Explainability: SHAP, LIME, ELI5

Data Engineering

Pandas • Polars • PySpark • Databricks

Cloud Infrastructure

  • Azure: ADF, Synapse, Databricks, Azure ML
  • AWS: S3, Athena, Redshift, ECR, SageMaker

MLOps & DevOps

  • Experiment Tracking: MLFlow
  • Deployment & Orchestration: Docker, Kubernetes, Databricks Asset Bundles
  • CI/CD: GitHub Actions, Azure DevOps
  • Quality & Monitoring: Great Expectations, Evidently
  • Development: Git, Unit Testing, Integration Testing

Development Tools

FastAPI • Flask • Streamlit • Bokeh • Panel

Domain Expertise

Core Machine Learning

Regression • Classification • Clustering • Feature Engineering • Model Selection

Time Series & Forecasting

Univariate & Multivariate Forecasting • Seasonality Modeling • Trend Analysis • Forecast Evaluation

Pattern Recognition & Anomaly Detection

Motif Discovery • Change Point Detection • Outlier Detection • Real-time Anomaly Detection

Causal Inference & Experimentation

Causal Analysis • A/B Testing • Treatment Effect Estimation • Counterfactual Reasoning • Experimental Design

Specialized Applications

  • Media Mix Modeling: Marketing Attribution, Channel Optimization, ROI Measurement
  • IoT Analytics: Sensor Data Processing, Predictive Maintenance, Equipment Monitoring
  • Entity Deduplication: Fuzzy Matching, Record Linkage, Master Data Management
  • LLM Applications: RAG Systems, Knowledge Retrieval, Semantic Search