Bruno Nogueira Carlos
Verified Expert in Engineering
ML Engineer and Developer
Medellín, Colombia
Toptal member since August 20, 2026
Bruno is a senior machine learning engineer and data scientist with over 10 years of experience in classical ML, computer vision, and production ML systems. He has delivered end‑to‑end workflows for clients such as GroupM and Mind Over Data, working on projects for Coca‑Cola, Bayer, Colgate-Palmolive, Universal Pictures, and Google‑partnered initiatives. His expertise spans Google Cloud Platform and Vertex AI.
Portfolio
Experience
- Feature Engineering - 10 years
- Exploratory Data Analysis - 10 years
- Data Preparation - 10 years
- Data Cleaning - 10 years
- EDA - 10 years
- Supervised Machine Learning - 4 years
- Computer Vision - 4 years
- Scikit-learn - 4 years
Preferred Environment
BigQuery, Google Cloud Storage, Cloud Run, Vertex AI, Python, Pandas, Scikit-learn, Multimodal Models, Embedding Models, Jupyter Notebook
The most amazing...
...experience has been delivering end‑to‑end classification systems using embeddings and robust tabular pipelines, deployed reliably in real‑world production.
Work Experience
Data Scientist
GroupM (WPP)
- Built production ML workflows spanning feature engineering, training, evaluation, batch inference, and real‑time inference for an LLM-based chatbot that queried BigQuery tables directly.
- Containerized services with Docker and deployed practical ML workloads through Cloud Run and simple Vertex AI endpoints.
- Developed computer vision and multimodal solutions with TensorFlow, including CNN-based approaches combining image, tabular, text, behavioral, historical, and metadata signals.
- Architected and refined classical ML models with Scikit-learn, XGBoost, and LightGBM for predictive, ranking, and business decision-support use cases.
- Collaborated on ML and analytics projects for global clients, including Coca‑Cola, Bayer, Colgate-Palmolive, Universal Pictures, and Google‑partnered initiatives with weekly technical reviews.
- Monitored model performance, stability, and drift, contributing to iterative model improvements and reliable production behavior.
- Reduced compute waste and improved prediction consistency by applying deduplication and noise-filtering techniques to high-volume visual datasets used by global clients.
- Leveraged pre‑trained TensorFlow models to accelerate model development, reducing experimentation cycles and delivering a production‑ready classifier faster.
Machine Learning Engineer
Empowerment Labs
- Served as technical lead in a newly created data science area, establishing analytics capabilities around customer knowledge, application usage, service monitoring, anomaly detection, and benchmarks.
- Developed product-oriented feature engineering and predictive modeling initiatives to support product, marketing, and operational decisions.
- Built end‑to‑end machine learning pipelines, delivering actionable insights and automated monitoring workflows that improved product reliability and decision‑making.
Machine Learning Engineer
PJBank
- Helped structure the data science area with SAS Viya in collaboration with data engineering teams and guided studies, priorities, and team efforts.
- Delivered churn, segmentation, cohort, fraud, credit-risk, and customer analytics, including a package of 18 real-time credit analysis models using machine learning and statistical techniques.
- Designed and deployed scalable scoring pipelines that powered real‑time credit decisions and improved operational efficiency across financial products.
Senior Data Scientist
Mind Over Data
- Developed customer knowledge and retail analytics solutions using statistical modeling, segmentation, propensity modeling, scoring, ranking, and behavioral analysis.
- Translated complex analytical problems into practical next-best-offer and decision-support solutions for diverse business needs.
- Engineered end‑to‑end analytical workflows that integrated customer behavior, transactional data, and business rules to support personalized recommendations and strategic decision‑making.
Experience
Plants Seedling Classification
The project involved image preprocessing, normalization, data augmentation, and the development of a convolutional neural network (CNN) architecture using TensorFlow and Keras. I experimented with both custom CNN models and transfer learning approaches to improve accuracy and robustness.
Model performance was evaluated using standard classification metrics, and hyperparameters were optimized to achieve better generalization on unseen images. The final solution demonstrates how deep learning and computer vision can be applied to real‑world agricultural challenges, enabling faster and more reliable seedling identification.
Credit Card Churn Prediction
To address the severe class imbalance in the dataset, I applied SMOTE and experimented with ensemble methods, including random forest, bagging, and boosting. I used cross‑validation and systematic hyperparameter tuning to improve model stability, reduce variance, and enhance generalization.
The final model delivered clear churn drivers and actionable insights for the operations team, enabling better customer retention strategies and more informed decision‑making. This project demonstrates my ability to design end‑to‑end ML pipelines for real business problems.
Personal Loan Campaign Modeling
The workflow included extensive exploratory data analysis, missing value treatment, outlier handling, and data preprocessing to ensure high‑quality inputs for the model. I built and evaluated decision tree models, applied pruning techniques to reduce overfitting, and compared performance across multiple configurations.
The final solution provided clear customer segments, actionable insights, and data‑driven recommendations that helped the business optimize marketing strategies and improve conversion rates. This project demonstrates my ability to build end‑to‑end ML solutions for practical business applications.
Support Ticket Categorization
The workflow included exploratory data analysis, text preprocessing, and experimentation with transformer‑based models. I designed and refined prompts to improve classification accuracy and ensure the model could interpret complex, real‑world ticket descriptions.
The final solution demonstrated strong performance in identifying ticket categories and extracting actionable information, reducing manual triage time and improving operational efficiency. This project highlights my ability to apply LLMs and prompt engineering to practical business problems.
Bank Customer Churn Prediction
The workflow included exploratory data analysis, data cleaning, normalization, and preparation of numerical and categorical features for neural network training. I implemented the ANN using TensorFlow and Keras, experimented with different architectures, activation functions, and optimization algorithms, and evaluated performance using standard classification metrics.
The final model provided clear churn indicators and actionable insights for the operations team, enabling more targeted interventions and improved customer retention. This project demonstrates my ability to design and train neural network models for real‑world business applications.
Education
Postgraduate Diploma in Artificial Intelligence and Machine Learning
University of Texas at Austin - Remote
Postgraduate Diploma in Marketing Research and CRM | Statistics and Information Management
Nova Information Management School (Nova IMS) - Lisbon, Portugal
Bachelor's Degree in Statistics
National School of Statistical Sciences - Rio de Janeiro, Brazil
Skills
Libraries/APIs
XGBoost, PySpark, TensorFlow, Scikit-learn, Pandas, NumPy, OpenCV, Hugging Face Transformers, CatBoost, PyTorch, Keras
Tools
BigQuery, AI Prompts, SQL Prompt, Microsoft Copilot, Visual Language Models (VLMs), ChatGPT, Claude, Git, Jupyter, You Only Look Once (YOLO), Codex
Languages
Python, SQL
Paradigms
Anomaly Detection, Synthetic Data Generation, Business Intelligence (BI), Object-oriented Programming (OOP)
Platforms
Vertex AI, Jupyter Notebook, Google Cloud Platform (GCP), Docker, Cloud Run, Databricks, Azure
Storage
Data Pipelines, MySQL, Google Cloud Storage, JSON, PostgreSQL
Frameworks
LightGBM
Other
Multimodal Models, Data Cleaning, Data Preparation, Predictive Modeling, Multivariate Analysis (MVA), Deep Learning, Feature Engineering, EDA, Computer Vision, Churn Analysis, Fraud Prevention, Supervised Machine Learning, Data Visualization, Random Forests, Cross-validation, Hyperparameter Tuning, Decision Trees, Data Preprocessing, Normalization, Data Science, Machine Learning, Data Analysis, Artificial Intelligence (AI), Model Evaluation, Data Scientist, Object Detection, Product Engineering, Applied AI, Data Engineering, Model Development, Monitoring, Linear Regression, Technical Documentation, Optical Character Recognition (OCR), Data Modeling, CSV, Convolutional Neural Networks (CNNs), Survey Design, Large Language Models (LLMs), Semantic Segmentation, RAG Pipelines, Unsupervised Machine Learning, Transfer Learning, Natural Language Processing (NLP), Retrieval-augmented Generation (RAG), Feasibility Studies, Image Processing, Hyperparameter Optimization, Data Labeling, Model Deployment, 2D Image Processing, Time Series Forecasting, Labeling, Kimi, Generative Artificial Intelligence (GenAI), APIs, Startups, Demand Forecasting, Combinatorial Optimization, Vertex, Prompt Engineering, Agentic AI, Dashboards, BI Reporting, Computer Vision Algorithms, AI Agents, API Integration, Excel 365, Microsoft 365, Embedding Models, Exploratory Data Analysis, Statistics, Statistical Modeling, Clustering, Supervised Learning, Unsupervised Learning, Sampling, Data Manipulation, CRM, Market Segmentation, Data Mining, Principal Component Analysis (PCA), LLM Fine-tuning, Word2Vec, Neural Networks, Word Embedding, Image Segmentation, Chatbots, ML Pipelines, Rankings, Know Your Customer (KYC), Transformers, Personally Identifiable Information (PII), Multimodal GenAI, Time Series, Geospatial Data, Geospatial Analytics, Causal Inference, Google Gemma, Hugging Face, Molmo, Qwen, Communication, Stakeholder Management, Azure Databricks, ChatGPT API, Dashboard Development, Reinforcement Learning
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