André Brandão

Senior Software Engineer, and Data Scientist

Seedz•  May 2023 - Present

Seedz merged with Gaivota, my previous employer. This company is dedicated to optimizing the agribusiness ecosystem. The main product is a framework to enhance sales campaigns in agribusiness, providing rewards for sellers and cashback for customers. I am leading a data monitoring team that supports Product, Business, and Operations teams in boosting the effectiveness of agribusiness companies. We perform Data Analysis, Process Mining, Machine Learning, Clustering, Data Visualization (e.g., Dashboards). We also develop Webhooks, REST APIs, and pipelines to extract, transform, and load data (ETL). Achievements: ■ REST API services to automate processes, reducing from up to 3 weeks to 2 hours the report generation ■ data lake to establish a centralized data hub to facilitate reporting and analysis across the organization ■ built solutions to uphold data quality throughout the ingestion process ■ conducted analysis to improve framework for sales campaigns to provide rewards for sellers and cashback for customers ■ dashboards that enable Operations teams to identify issues in campaigns, such as reward failures Tools: AWS, Azure DevOps, Docker, Git, Python, Pandas, Spark, Statistics, Matplotlib, Plotly, Metabase, Scikit-Learn, Jupyter, Flask, Swagger, AsyncIO, Celery, MySQL, Postgres, Redshift, MongoDB, DynamoDB, Redis, SQS, EC2, ECS, CloudWatch, S3, Lambda

Senior Data Scientist, and Manager

Gaivota•  April 2021 - April 2023

Leading a data science team that supports Product and Business teams in boosting the effectiveness of agribusiness companies. We performed Data Analysis, Process Mining, Machine Learning, Clustering, Data Visualization (e.g., Dashboards). We also developed Webhooks, REST APIs, and pipelines to extract, transform, and load data. Achievements: ■ web crawlers to systematically collect public data from governmental organizations, property listings, stock exchange, and others ■ a machine-learning model utilizing Deep Learning techniques to predict soybean yields to well-known cities in Brazil for its exceptional soybean productivity ■ dashboards reporting insights to assist banks and tradings in compliance analysis and business plans ■ REST API service to deliver property valuations based on location. To achieve this, we implemented a web crawler to systematically gather pricing and property features data. Following data collection, we conducted in-depth data analysis and developed a robust machine-learning model that produced highly accurate results. As a result, our solution drastically cut the property valuation time from 8-10 days to a few minutes and successfully integrated into a major Brazilian bank's credit provision pipeline for growers ■ statistical, process mining methods, and dashboards that enabled Product teams to identify user interaction patterns and preferences, event flows, pinpointed bottlenecks, underutilized pages, and clustered users based on their behavioral patterns. This initiative significantly enhanced the user experience and contributed to improved user retention on the platform Tools: AWS, Azure Devops, Jenkins, Argo CD, Docker, Git, Python, Pandas, Spark, Numpy, Pm4py, Statistics, Matplotlib, Plotly, Metabase, Scrapy, Scikit-Learn, Jupyter, Streamlit, FastAPI, Tornado, Flask, Swagger, AsyncIO, Pika, MySQL, Postgres, MongoDB, RabbitMQ, EC2, ECS, CloudWatch, S3, Lambda, Athena, Glue, DMS, Lake Formation, EMR

Senior Software Engineer, and Data Scientist

ExACTa PUC-Rio•  April 2020 - April 2021

Conducted data analysis and exploration of data from Petrobras, a major Oil and Gas company in Brazil, focused on identifying patterns, and developing prediction models and dashboards to enhance operational efficiency. Achievements: ■ The company faced an issue of inaccuracies in reports of equipment inspection performed by experts. Our goal was to build an artificial intelligence model to assist inspectors in report writing. We constructed a comprehensive database of inspection reports and performed data analysis. Then, we developed a web service using a machine learning model to detect inaccuracies or gaps in reports and suggest appropriate terms for damages, causes, mechanisms, and actions. As a result, the harmonic mean of accuracy and precision (F1 score) reached a result greater than 95%. This project received the Petrobras Inventor Award 2022 in recognition of the results achieved and the impact on the efficiency of equipment maintenance in refineries. ■ Petrobras faced a critical issue with elevated hydrogen sulfide (H2S) emissions in its oil refineries, posing a severe threat to the safety of the local community. To assist the refineries in identifying and mitigating the root causes of elevated H2S emissions. We conducted in-depth data exploration, correlating refinery historical data with meteorological information. Subsequently, we developed a robust machine-learning model to provide prescriptive insights into H2S emissions. A Power BI dashboard was also created to visualize the factors influencing H2S levels every 5 minutes. Our solution encompassing machine learning and a dashboard led to a substantial reduction in H2S emissions and was successfully deployed across multiple refineries. Tools: Azure, Azure DevOps, Azure Functions, Docker, Git, Python, Pandas, Numpy, Statistics, Matplotlib, Seaborn, Plotly, Scikit-Learn, PyTorch, Jupyter, Flask, Swagger, MySQL, SQL Server, MongoDB, Git, Power BiConducted

Data Scientist

ECOA PUC-Rio•  October 2018 - March 2020

Analysis and exploration of data from Public Prosecution and Court of Law of Rio de Janeiro to increase the efficiency of the legal processes. I was responsible for conducting Data Analysis, Data Visualization, and Machine Learning modeling as natural language processing (NLP) to identify patterns in the court proceedings. Achievements ■ Prosecutors faced challenges in identifying legal texts with similar contextual content within the vast repository of historical documents. Our goal was to develop a service that empowers prosecutors by providing efficient assistance in identifying legal texts with similar content, enhancing their workflow. We conducted a data analysis of the historical documents from the Public Prosecution, employing data analysis techniques to extract and explore information from PDF files. Utilizing a pre-trained model with word embeddings, we built a service that retrieves similar texts through vector representations, employing cosine distance calculations and sorting the results accordingly. Our solution significantly improved the efficiency of document retrieval for prosecutors. It was successfully deployed enhancing the overall workflow and effectiveness of legal text analysis ■ I led a significant project to enhance the operational efficiency of the Public Prosecution in Rio de Janeiro by optimizing the allocation of prosecutors to individual legal cases. This involved analyzing the legal journey of lawsuits from the Court of Law to the Public Prosecution. After cleaning the data and identifying correlations within the dataset, we developed a machine learning model that could automate the matching of the most suitable prosecutor to each lawsuit. We then deployed this model as a service, seamlessly integrating it into the Public Prosecution's workflow and improving the lawsuit allocation Tools: Docker, Git, Python, Pandas, NumPy, Statistics, Scikit-Learn, Seaborn, Matplotlib, Plotly, Tableau, Flask, AsyncIO, Jupyter, MySQL, MongoDB

Software Engineer, and Researcher

TeleMídia Lab. PUC-Rio•  August 2016 - October 2018

Working on the official Ginga Middleware development project, which is the middleware of the Japanese-Brazilian Digital TV System (ISDB-TB) and ITU-T H.761 recommendation for iPTV services. Achievements: ■ fixed reported bugs, and refactored the source code to improve its efficiency and memory usage ■ improved the video player used by Ginga to support operations of resize and freezing screen, setting audio, etc. Tools: C, C++, Lua, GStreamer, NCL, GDB, Git, XML, Shell script, JavaScript, MySQL, MongoDBWorking on the official Ginga Middleware development project, which is the middleware of the Japanese-Brazilian Digital TV System (ISDB-TB) and ITU-T H.761 recommendation for iPTV services. Achievements: ■ fixed reported bugs, and refactored the source code to improve its efficiency and memory usage ■ improved the video player used by Ginga to support operations of resize and freezing screen, setting audio, etc. Tools: C, C++, Lua, GStreamer, NCL, GDB, Git, XML, Shell script, JavaScript, MySQL, MongoDB Skills: Linux · GNU Debugger · XML · NCL · Git · Lua · JavaScript · C (Programming Language) · REST APIs · Shell script · SQL · Machine Learning · MySQL · Software Development · MongoDB · C++ · Gstreamer

Software Engineer, and Founder

Mediabox Technologies•  December 2014 - January 2017

Leading a team responsible for developing an authoring tool for multimedia Learning Objects (LOs) to allow teachers to create multimedia educational content for interactive TV and the Web without requiring programming skills. Achievements ■ developed Cacuriá, a desktop tool to support instructors in creating multimedia educational content for TV and Web, used by 13 universities ■ integrated Cacuriá with the iVoD (Interactive Video on Demand) service from RNP - a National Research and Educational Network responsible for promoting the development of networks in Brazil ■ launched Mestrar, an online platform to store and deliver Learning Objects created by Cacuriá ■ conducted courses to disseminate the Cacuriá in Brazil. Tools: C, C++, Qt, Lua, NCL, GDB, Git, XML, Shell script, PHP, JavaScript, MySQLLeading a team responsible for developing an authoring tool for multimedia Learning Objects (LOs) to allow teachers to create multimedia educational content for interactive TV and the Web without requiring programming skills. Achievements ■ developed Cacuriá, a desktop tool to support instructors in creating multimedia educational content for TV and Web, used by 13 universities ■ integrated Cacuriá with the iVoD (Interactive Video on Demand) service from RNP - a National Research and Educational Network responsible for promoting the development of networks in Brazil ■ launched Mestrar, an online platform to store and deliver Learning Objects created by Cacuriá ■ conducted courses to disseminate the Cacuriá in Brazil. Tools: C, C++, Qt, Lua, NCL, GDB, Git, XML, Shell script, PHP, JavaScript, MySQL Skills: Linux · XML · NCL · Git · Lua · JavaScript · C (Programming Language) · REST APIs · Qt · SQL · MySQL · Project Management · Software Development · C++

Researcher, and Software Engineer

Laboratory of Advanced Web Systems - LAWS/UFMA•  December 2011 - April 2015

Working on the Cacuriá development project, an authoring tool for multimedia Learning Objects (LOs) to allow teachers to create multimedia educational content for interactive TV and the Web without requiring programming skills. Cacuriá is integrated with the iVoD (Interactive Video on Demand) service from RNP, a National Research and Educational Network responsible for promoting the development of networks in Brazil, including the development of innovative applications and services.Working on the Cacuriá development project, an authoring tool for multimedia Learning Objects (LOs) to allow teachers to create multimedia educational content for interactive TV and the Web without requiring programming skills. Cacuriá is integrated with the iVoD (Interactive Video on Demand) service from RNP, a National Research and Educational Network responsible for promoting the development of networks in Brazil, including the development of innovative applications and services. Skills: Linux · XML · NCL · Git · Lua · JavaScript · C (Programming Language) · PHP · Qt · SQL · Software Development · NoSQL · C++

Software Engineer

BR Ativos•  March 2012 - February 2013

Working on the development and support of a system to manage municipal tributes used by 17 cities from Maranhão, Brazil.Working on the development and support of a system to manage municipal tributes used by 17 cities from Maranhão, Brazil. Skills: Linux · HTML · Git · JavaScript · PHP · SQL · MySQL · Software Development

Back-end Developer

HCG Engenharia de Sistema Ltda•  June 2011 - March 2012

Working on the development and support of a system to manage municipal tributes used by Brazilian cities.Working on the development and support of a system to manage municipal tributes used by Brazilian cities. Skills: HTML · Git · SQL · Software Development