Cleyton Pires
Verified Expert in Engineering
Data Scientist and Developer
Florianópolis - State of Santa Catarina, Brazil
Toptal member since November 18, 2022
Cleyton is a software engineer with 20+ years of experience in all aspects of the software development lifecycle, from concept to development, testing, and management. A specialist in data science and machine learning, he is Google certified and well-versed in various technologies, including Python, Jupyter Notebook, pandas, NumPy, and web scraping. Cleyton is also skilled in back-end technologies, such as Flask, RESTful API, authentication, SQLAlchemy, and PostgreSQL.
Portfolio
Experience
- Software Development - 20 years
- Python - 10 years
- SQL - 6 years
- Data Science - 5 years
- Back-end - 5 years
- Java - 4 years
- Machine Learning - 3 years
- Google Cloud - 1 year
Work Experience
Lead Data Scientist
Turing
- Led a team of 30–50 data scientists delivering training data and evaluation pipelines for a frontier, large-scale AI system developed by a top-tier technology company, ensuring high-quality outputs and alignment with strict engineering standards.
- Developed and reviewed high-quality Python notebooks and annotation guidelines to support supervised fine-tuning (SFT), improving the correctness, consistency, and testability of AI-generated code and model outputs.
- Conducted large-scale analysis of model outputs, identifying systematic failure patterns and proposing data-centric improvements (SFT and RLHF), significantly increasing robustness and real-world reliability.
- Designed and implemented a simulation framework with realistic, mocked service APIs (e.g., calendar, eCommerce, social platforms), enabling end-to-end validation of agent behavior through scenario-based testing and state assertions.
- Built an automated evaluation system for multi-turn interactions using LLM-driven user simulation, uncovering edge cases and validating the correctness and completeness of agent-generated actions and test scenarios.
- Benchmarked model performance against public evaluation suites and internal tasks, identifying gaps in reasoning, code execution, and test coverage, and driving targeted improvements through curated training data.
Data Scientist (Tax Officer)
Federal Revenue Service of Brazil
- Led the development of the Brazilian government website "Portal Siscomex" (portalunico.siscomex.gov.br), aimed to reduce bureaucracy costs and increase the efficiency of imports and exports.
- Built a machine learning model based on supervised learning to identify signs of tax evasion from individual taxpayers.
- Led the restructuring of the tax collection division, which comprised 230 members. Local teams were transformed into seven remote teams specialized by subject, resulting in a 20% increase in productivity in the first three months of operation.
- Contributed to audit case selection and taxpayer compliance, mainly using Python and SQL (Big Data) to apply data mining techniques to better understand, process, and conclude vast amounts of information.
Senior Software Engineer
Ericsson
- Developed new features for the user equipment handling (UEH) subsystem of Ericsson's radio network controller (RNC) using the C++ programming language.
- Tracked and fixed bugs using a multi-level logging strategy.
- Created the software design specifications for new features, including VoIP and RAB modification.
Test Lead
Siemens
- Led outsourced test teams consisting of 2-6 members. Participated in several projects with suppliers, including AppMenu for the Siemens cell phone main menu, JSR 238, and JSR 172.
- Developed unit testing using the JUnit/J2MEUnit framework.
- Applied the Scrum framework for agile software management and development.
- Defined and deployed testing strategies for each project.
Software Engineer
Motorola Mobility
- Coordinated and implemented a six-month project to port the GSM messaging component to the Motorola 3G platform, regarded as one of the greatest achievements of Motorola R&D Brazil in 2004.
- Developed messaging features such as SMS, EMS, and MMS for Motorola mobile phones using the C programming language in a Unix environment.
- Tracked and fixed bugs related to messaging features.
- Used computer-aided software engineering (CASE) tools for software development, including Rational Rose, ClearCase for version control, and ClearDDTS for bug tracking.
Experience
Government Website
Secrets Web App
https://github.com/cleytonap/secretsIt is a simple authentication app with user login, logout, registration, and access control to be a starting point for other more complex Node.js applications. It supports two Passport.js strategies for user authentication, including local (username and password) and Google (OAut2.0). Data is persisted into the MongoDB database using Mongoose.
Containerized RESTful API (Demo)
https://github.com/cleytonap/wiki-apiContinuous Deployment from Git using Google Cloud Build (Demo)
https://github.com/cleytonap/gcp-webserviceBuild Type: Google Cloud Buildpacks (rather than Dockerfile).
Education
Postgraduate Degree in Data Science
Pontifical Catholic University of Minas Gerais (PUC-Minas) - Belo Horizonte, Minas Gerais, Brazil
Bachelor's Degree in Law and Justice Administration
Federal University of Mato Grosso (UFMT) - Cuiabá, Mato Grosso, Brazil
Bachelor's Degree in Computer Science
Federal University of Santa Catarina (UFSC) - Florianópolis, SC, Brazil
Certifications
Google Cloud Certified Professional Machine Learning Engineer
Sun Certified Java Programmer
Sun Microsystems
Skills
Libraries/APIs
REST APIs, EJS, Node.js, Passport.js
Tools
Microsoft Excel, Git, Apache Impala, Google AI Platform, BigQuery, Mongoose
Languages
Python, C++, SQL, Java, Excel VBA, C, JavaScript, CSS
Platforms
Jupyter Notebook, Visual Studio Code (VS Code), Unix, J2ME
Paradigms
Agile Project Management, Unit Testing, Testing, Test Automation, Scrum, Functional Programming, Automated Testing, Design Patterns, Continuous Deployment
Storage
Databases, PostgreSQL, Data Lakes, Google Cloud, MongoDB
Frameworks
OAuth 2, Flask
Other
Software Development, Algorithms, Programming, Scripting, Software Design, Version Control Systems, Bug Fixes, Code Review, High Code Quality, Large Language Models (LLMs), AI Model Training, Data Science, Machine Learning, Data Mining, Workflow Automation & System Integration, Remote Team Leadership, Team Leadership, Back-end, Big Data, Google Cloud ML, Data Engineering, Web Development, Web Services, Containers, Agentic AI, Model Evaluation, Data Annotation, Log Analysis
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