Damage Risk on Gas Pipelines
At Keyrus, I developed an innovative model to support the field engineering team in making more assertive decisions about which street works should be prioritized for daily inspections.
The main challenge was to identify the works with the highest likelihood of causing damage to the city’s underground gas pipelines, while optimizing the technicians’ logistics given limited resources. Damage to the pipelines could result in fatal explosions and significant material losses.
The project led to a significant reduction in network damages, as well as a considerable increase in both the productivity and quality of the engineers’ work.
Photovoltaic Power Forecast
As electricity consumption grows rapidly worldwide, renewable energy resources, such as solar power, play a crucial role in this scenario, contributing to meet the demand in a sustainable manner. Although the participation of photovoltaic (PV) power generation has increased in recent years, PV systems are quite sensitive to weather and climatic conditions, leading to an undesirable variability in power production.
To improve the stability, reliability, and management of the electrical grid, accurate forecasting models that relate operational conditions to power production are necessary. In this work, we evaluate the performance of regression methods applied to short-term (next-day) forecasting of the power production of a photovoltaic plant. Specifically, we consider five regression methods and different feature set configurations.
Our results suggest that MLP and SVR provide the best forecasting results, overall. Additionally, while features based on different levels of solar irradiance play a fundamental role in predicting power generation, the use of additional features can improve the forecasting results.
Predictive Maintenance & Anomaly Detection
At Shape Digital, my primary focus is on developing advanced predictive maintenance models using intricate spindle data. This involves leveraging machine learning to anticipate equipment failures, minimize downtime, and optimize operational efficiency for industrial clients.
My contributions extend to critical projects such as anomaly detection in offshore sensor data, ensuring the integrity and safety of remote operations. I have also pioneered corrosion detection solutions using computer vision, providing proactive measures against structural damage.
Currently, I am developing sophisticated models for process safety and risk assessment tailored for offshore plants. These initiatives are crucial for enhancing operational safety and mitigating potential hazards in high-stakes environments.
Professional Journey: Keyrus
Data Scientist
May 2022 - July 2024 (2 years 3 months)
São Paulo, Brazil
Successfully delivered projects related to damaged prediction on gas distribution network, billing rule optimization, and the application of People Analytics to derive meaningful insights and drive informed decision-making.
Professional Journey: Grupo GCB & THI
Grupo GCB - Data Scientist
September 2021 - May 2022 (9 months)
- Performed extensive exploratory data analysis (EDA).
- Focused on creating consistent databases and advanced feature engineering.
- Developed and improved ML models for predicting credit-related losses and monitoring them in production.
- Deployed end-to-end predictive models for prospecting new customers based on credit risk and score.
Technische Hochschule Ingolstadt - Internship Researcher
September 2020 - February 2021 (6 months)
Ingolstadt, Bavaria, Germany
- Developed a machine learning approach to accelerate occupant safety simulation by replacing time-consuming Finite Element (FE) simulations.
Universidade Federal de Santa Catarina - Graduate Teaching Assistant
August 2019 - December 2019 (5 months)
Joinville, Brazil
- Assisted classes and evaluations.
- Taught topics including Buses and Communication Protocols, Standard Two-Wire Serial Interface, and Serial Peripheral Interface.
Publications & Key Competencies
Selected Publications
- "Rapid Estimation of Occupant Crash Behavior Considering Human Diversity"
- "Comparative Study of Photovoltaic Power Forecasting Methods"
These publications highlight my research capabilities and contributions to both engineering and data science fields, showcasing my ability to translate complex theoretical concepts into practical applications.
Key Competencies
AI & Machine Learning
Model Development, Deployment, MLOps
Data Science
EDA, Feature Engineering, Predictive Modeling, Risk Analysis
Big Data Engineering
PySpark, Databricks, Data Lakehouse Architecture
Cloud & DevOps
Azure, AWS, CI/CD Pipelines, Infrastructure as Code
My comprehensive skill set positions me as a versatile professional ready to tackle cutting-edge challenges in AI, ML, and data engineering.