Stéphane Thibaud
Verified Expert in Engineering
Data Scientist and Developer
Lisbon, Portugal
Toptal member since February 10, 2026
Stéphane is a data scientist and software engineer with 12+ years of professional experience. Over the past five years, he has focused on applied machine learning and statistical estimation, building rigorous models that turn complex, real-world data into reliable decision support. His experience spans forecasting, estimation, classification, and unstructured data, with a strong emphasis on experimental design and robust evaluation.
Portfolio
Experience
- Python - 13 years
- SQL - 13 years
- Software Engineering - 13 years
- Programming - 13 years
- Scikit-optimize - 5 years
- Scikit-learn - 5 years
- Pandas - 5 years
- Statistics - 5 years
Preferred Environment
Python, Pandas, NumPy, SciPy, Scikit-learn, Scikit-optimize, SQL, FastAPI, Agentic Coding, Haskell, Claude API
The most amazing...
...project I've worked on was designing a fraud detection model that uncovered real criminal transactions with strong precision.
Work Experience
Data Scientist
Rakuten
- Co-designed and tuned a retrieval-augmented generation (RAG) chatbot for internal operations; optimized chunking and embedding-based retrieval (cosine similarity). Awarded a company-wide Grand Prize (February 2024) for impact.
- Developed and deployed a time-to-conversion prediction system enabling targeted lifecycle actions.
- Implemented and adapted a published textual-gradient prompt optimization method for chatlog classification, improving performance by around 20% MCC (relative).
- Built a causal budget allocator for marketing spend using counterfactual methods; achieved measurable reduction in cost-per-acquisition under budget reallocation experiments.
- Contributed to a predictive store location revenue model supporting data-driven site selection decisions.
- Mentored two colleagues, supporting onboarding and providing technical guidance on modeling and experimentation.
Data Scientist
Caulis
- Developed anomaly-detection analyses to surface high-risk transactions and prioritize investigations.
- Mined candidate fraud-detection rules from historical data to support rule design and refinement.
- Designed graph-based visualizations of transaction flows for internal analysis and discussion.
- Implemented automated anomaly notifications via email to flag unusual financial activity.
Python Developer
Merrill
- Extended a financial analytics platform with new functionality.
- Modernized a legacy Python codebase from Python 2 to Python 3.
- Improved test coverage and system performance across core components.
Software Engineer
Drivemode
- Designed and executed a zero-downtime migration from a relational database to a managed cloud NoSQL datastore.
- Developed a robust, extensible data export framework supporting multiple back-end sources and targets.
- Integrated and supported production deployment of a churn-prediction model, including data pipeline integration and operational monitoring.
Software Engineer
KE-chain
- Contributed to the development of a web-based engineering software platform.
- Implemented and improved platform capabilities related to versioning and system performance.
- Integrated the platform with external systems using standardized interfaces.
Experience
Employee Stock Option Valuation Tool
https://eso.stephanethibaud.xyz/The tool emphasizes correctness, transparency of assumptions, and practical decision support over purely theoretical pricing. The online interface allows users to interactively vary model parameters and visualize how each assumption affects the resulting option value. The project is MIT-licensed, publicly available on GitHub, and designed to be reusable in analytical workflows or decision-support contexts.
Explainable Fraud Detection with Temporal Validation (Public PaySim Dataset)
https://www.kaggle.com/code/snthibaud/fraud-detectionKey elements include feature engineering (log-transformed balances and relative transaction ratios), removal of high-cardinality identifiers to improve generalization, class-weighted decision trees, and randomized hyperparameter search.
Model performance was evaluated using ROC-AUC with cross-validation and a final time-based split to simulate forward-looking fraud prediction. The final model achieved very high AUC on the hold-out set and remains fully interpretable via decision-tree inspection.
Education
Master of Science Degree in Software Engineering
Open University of The Netherlands (OU) - The Netherlands
Bachelor's Degree in Computer Science
Eindhoven University of Technology (TU/e) - Eindhoven, The Netherlands
Certifications
Certified Yoga Teacher (200-hour Program Meeting RYT-200 Standards)
Yogaworks
QTM4x: Fundamentals of Quantum Information
DelftX
QUAN11000: Quantum Computing for Everyone 1
UChicagoX
EEX0001B: Quantum Mechanics for Scientists and Engineers 2
StanfordOnline
SOE-YEEQMSE01: Quantum Mechanics for Scientists and Engineers 1
StanfordOnline
Japanese Language Proficiency Test N2
Japan Educational Exchanges and Services and the Japan Foundation
8.01.1x: Mechanics: Kinematics and Dynamics
MITx
CSMM.101x: Artificial Intelligence (AI)
ColumbiaX
Skills
Libraries/APIs
Pandas, Scikit-learn, OpenAI API, Claude API, NumPy, SciPy, Scikit-optimize, NetworkX, XGBoost, SpaCy, PySpark, Pydantic, Python API, REST APIs, LSTM
Tools
ChatGPT, GitHub, Seaborn, Claude, Pytest, AI Prompts, Git, BigQuery, StatsModels, Claude Code
Languages
Python, SQL, Regex, Scala, Kotlin, Java, C++, R, JavaScript, Haskell
Frameworks
Multi-armed Bandits (MABs), LightGBM, Streamlit, Django, Ext JS, LangGraph, LlamaIndex, Jinja, Agentic Frameworks, Yoga
Paradigms
Conversion Rate Optimization (CRO), Functional Programming, Software Testing, Design Patterns, Security Software Development, Anomaly Detection, ETL, Microservices, Continuous Integration (CI), Continuous Delivery (CD)
Platforms
Google Cloud Platform (GCP), Ubuntu, Amazon Web Services (AWS), Docker, Firebase, AWS Lambda, LangSmith, Kubernetes, Visual Studio Code (VS Code)
Storage
Apache Parquet, Cloud Firestore, Data Validation, Google Cloud Storage, Databases, Elasticsearch, Google Cloud Datastore, MySQL, MongoDB, NoSQL, PostgreSQL, Data Pipelines, Amazon S3 (AWS S3)
Industry Expertise
Applied Statistics, Marketing
Other
Random Forests, Statistics, Programming, Algorithms, Software Engineering, Artificial Intelligence (AI), Combinatorial Optimization, Large Language Models (LLMs), Retrieval-augmented Generation (RAG), Machine Learning, Data Science, Time Series Forecasting, Dashboards, Data Analysis, Forecasting, Model Validation, ChatGPT API, Generative Artificial Intelligence (GenAI), Analytics, Data Analytics, Sequence Models, Time Series Data, Word Embedding, RAG Systems, Semantic Similarity (Cosine Similarity), Text Classification, Prompt Optimization, Data Extraction, Classification, Feature Engineering, AI Model Training, Model Evaluation, AI Assistants, OpenAI, Scientific Data Analysis, Cohort Analysis, Hypothesis Testing, Time Series Analysis, Regression Modeling, Predictive Analytics, System Modeling, Attribution Modeling, Customer Segmentation, Funnel Marketing, Markov Model, Marketing Analytics, Marketing Attribution, Parquet, Quantitative Analysis, Incrementality Testing, Funnel Analysis, ML Pipelines, Cost Reduction & Optimization (Cost-down), Source Code Review, System Architecture, PDF, API Integration, Cloud Architecture, AI Chatbots, Prompt Engineering, Architecture, Data Scientist, Statistical Analysis, Chatbots, Quantitative Modeling, Multivariate Statistical Modeling, zero shot learning, Data Augmentation, Statistical Methods, Financial Modeling, Cloud Platforms, Mathematical Modeling, Full-stack Development, Back-end, Linear Regression, Dimensionality Reduction, Gradient Boosting, K-means Clustering, Logistic Regression, Time Series, Data Integrity, Pattern Recognition, AI Automation, ChatGPT Prompts, Datasets, Documentation, Risk Models, Data Preprocessing, Large Data Sets, Model Deployment, Model Development, JupyterLab, Discrete Mathematics, Computer Graphics, Computer Systems, Computer Networking, Security, International Negotiation, Law, Logic and Set Theory, Calculus, Linear Algebra, Human-Technology Interaction, Data Structures, Programming Methods, Automata and Process Theory, Macro-Economics, Operating Systems, Software Specification, Probability Theory, Distributed Algorithms, Business Information Systems, Datamining and Knowledge Systems, Web Technology, Software Development Management, Software Composition, Software Architecture, Research, Writing & Editing, Software Evolution, Software Verification and Validation, Causal Inference, Multi-Objective Optimization, Genetic Algorithms, Derivative Pricing, Mechanics, Kinematics, Japanese, Quantum Mechanics, Quantum Computing, Google Cloud Dataflow, Data Modeling, Deep Learning, Spatial Analysis, LangChain, Statistical Modeling, eCommerce, ETL Pipelines, Natural Language Processing (NLP), RAG Pipelines, Vector Databases, Agentic RAG Systems, Financial Data, Data Engineering, Embedding Models, Scalable Vector Databases, Decision Trees, Hyperparameter Tuning, Data Classification, Model Interpretability, Data Visualization, Data Handling, Trading, Ethics, Customer Journey, Document Parsing, AI Architecture, Bayesian Inference & Modeling, Quantitative Finance, Full-stack, Large Language Model Operations (LLMOps), RAG Architecture, Reporting, Revenue Optimization, Reinforcement Learning, Chatbot Conversation Design, Technical Leadership, Financial Markets, Fixed Income, Dashboard Development, Transformer Models, Deep Reinforcement Learning, APIs, Google BigQuery, Bayesian Machine Learning, Bayesian Statistics, Front-end, Startups, Data Quality, Support Vector Machines (SVM), FastAPI, API Gateways, Anthropic, CI/CD Pipelines, Distributed Systems, Data Migration, Agentic AI, Monitoring, Telemetry, Machine Learning Operations (MLOps), SDKs, AI Agents, Finance, Communication, Coaching, Stakeholder Management, Agentic Coding
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring