Work experience

Aravita 

Senior Data Scientist

1. Operations research in perishable items: Responsable for developing state-of-the-art machine learning and optimization methods to deal with the pershable items product ordering

2. Demand forecasting of perishable items: Developing state-of-the-art machine learning method for demand forecasting that encompass feature engineering and model tunning of internally developed method.

Data Marketplace

Senior Data Scientist

1. Data Qualification and Fact Checking: Leveraged LLM - ChatGPT and other open-source methods to implement applications for data qualification and fact-checking for audits. This initiative enhanced the accuracy and efficiency of the auditing process.

2. Anomaly Detection: Developed an anomaly detection system using Natural Language Processing (NLP) and clustering techniques to identify unnecessary procedures or materials in client health accounts. This system expedited the process of identifying unnecessary expenses, thereby saving time and resources.

3. Expense Forecasting: Implemented a machine learning tool to predict the occurrence of unnecessary expenses in the health system. The tool was developed by comparing various forecasting techniques, including ARIMA, Prophet, LSTM, MLP, among others, to ensure optimal performance.

3. System Migration and Re-implementation: Led the migration of the existing system to a cloud platform and re-implemented several machine learning algorithms in the Databricks pipeline using PySpark. This process was adjusted to align with Azure MLflow and DevOps, enhancing system efficiency and scalability.

Polytechnic of Turin - Department of Mathematical Sciences

Machine Learning Researcher

1. Double Ph.D. agreement between University of Sao Paulo and Polytechnic of Turin.

2.  Reinforcement learning approach for the discrete lot-sizing problem (DLSP): with sequentially dependent setup cost and setup times greater than one time period. Using policy iteration methods (dynamic programming) and reinforcement learning, we solved the problem and tested both techniques for multiple items and machines.

3.  Reinforcement learning approach to solve the replenishment of perishable items: Here we compare with a linear approach that directly approximates the policy. Here, consumer dynamics are relevant to the problem and may generate non-linear characteristics that are not solvable by the linear approach we compare.

Polytechnic of São Paolo - University of São Paolo

Machine Learning Researcher

1. Double Ph.D. agreement between the University of Sao Paulo and Polytechnic of Turin.

2. Supervised learning  for finantial trading heuristic: We proposed a supervised learning approach using classification methods and state-of-the-art artificial neural networks to compare against reinforcement learning methods to solve decision problems in finance. We show the advantages of supervised learning over RL in asset trading, focusing on the single asset trading problem.

3. Sentiment analysis for reinforcement learning trading systems: We employ sentiment analysis as a feature extraction of news to improve the performance of financial trading systems employing reinforcement learning.

4. Synthetic and finantial time series forecasting: We explored forecasting techniques using synthetic generate time series. Considering each neural network architecture and time series, we show evidence that some better-fitted techniques for generated time series can perform better in real-world data such as financial time series.

ItaúBBA

Product Manager

1. Cash Management Services: Managed a portfolio of cash management services, including corporate credit cards, checking accounts, and cash transportation. Oversaw all stages of product lifecycle, from conceptualization to market launch and performance tracking.

2. Business Intelligence Application: Utilized business intelligence tools to drive product development efforts. This data-driven approach allowed for strategic focus on enhancing specific features of the products, leading to improved customer satisfaction and increased market share.

3. Blockchain Research: Conducted extensive research and exploration into blockchain technology. This initiative aimed to identify potential applications of blockchain in our product offerings and to stay ahead of emerging trends in the financial services industry.

Itaú-Unibanco

Software Engineer

1. Liaison with TCS and Totvs: You will have direct contact with prominent software consultancies, including Tata Consultancy Services (TCS) and Totvs. This involves facilitating robust communication channels, ensuring software requirements are clearly understood and fulfilled by both parties.

2. Software Migration for Custody Systems: You will be tasked with coordinating the software migration for custody systems. This will include planning, execution, and monitoring of migration processes, troubleshooting any issues that arise, and ensuring minimal system downtime.

3. Feature Specification and Implementation: A significant part of your role will involve the identification of essential features, their specification, and subsequent implementation within the software. This will require a deep understanding of user needs and industry trends to ensure the software's relevance and competitiveness.