Rodolfo Rodrigues Melo

Development of a tax compliance verification system to analyze digital accounting files for taxpayer compliance, potentially uncovering millions in irregularities.

The project involves the entire pipeline of establishing a tax compliance verification system. Currently, we are downloading TXT files representing Digital Accounting Files (ECD). A Java code has been developed to read and store these files in our Oracle databases. For data analysis, A Spark code was created to retrieve ECD information and to combine it with other data from our internal Data Warehouse and public company information from the Federal Revenue Service. This integration facilitates a comprehensive data cross-checking process aimed at detecting discrepancies. This approach has enabled us to identify Declaration Omissions by establishing a mandatory rule and checking taxpayer compliance. It also helps in pinpointing irregularities in analytical accounts. We anticipate that this extensive cross-referencing with the state’s major taxpayers will uncover a large amount of irregularities.

Development of a model for estimating fair market prices of products for accurate tax enforcement using Spark and NLP

This project aims to accurately identify and price products like soft drinks for tax purposes, especially for goods that are under-declared or under-invoiced at state borders. Utilizing methodologies from EloGroup and employing Spark for data processing, the project tackles the challenge of non-standardized electronic fiscal documents. My contribution is focused on refining the analytics and mathematics, using NLP in Spark, to improve product identification and price estimation, thus aiding in effective tax regulation.

A list of projects developed in the last 8 years. For more information, reach me out by email.

• State Processing Exctract: Developed by the Planning Audit team innovative, this work created a new validation framework for the Brazilian IVA Declaration. In 2022, this project got the 2nd place in the Tribuatare Award, a innovative competition among the best tax audit projects in the country.
• Shell Company Detection: Using Graph Neural Networks to Detect Shell Companies.
• Medicine Price Reference: Methodology using NLP and Statistics to create price references for purchasing medications in the State. (Laboratory of Statistics and Data Science at UFAL, in collaboration with SEFAZ/PI).
• Tax Compliance Study: Using big data to eases the process through which small firms comply with tax regulation. (Inter-American Development Bank, in collaboration with SEFAZ/PI).
• Tax Gap Methodology: Improved the methodology from Nelson Leitão Paes (University of Brasilia - BR) using Eletronic Invoices.
• Tax Compliance Verifications: Implemented over 30 different rules to detect tax fraud / irregularities. A large amount of resources were recovered with these rules.

Leading Analytics Projects

• Consultant on Data Science projects for the Treasury Secretary
• Developed analytics solutions to establish market reference prices using Electronic Invoice data.
• Implemented automated systems to detect discrepancies in Digital Accounting Records.
• Collaborated with the Data Science Laboratory at the Federal University of Alagoas to create a reference pricing methodology for pharmaceuticals, enhancing procurement efficiency for the State of Piauí.

Work on the leadership of Professor Vincenzo Galasso in the Department of Social and Political Sciences

• Specialized in emotion detection through audio and textual analysis.
• Conducted opinion poll analysis to gauge public sentiment.
• Employed text mining techniques for extracting key insights from interviews.
• Simulated statistical models for forecasting voting trends in political acronyms.

Team Lead in the Auditing Unit

• Led a multidisciplinary team comprising tax auditors, data engineers, and analysts.
• Managed a real-time fraud detection initiative to enhance auditing efficiency.
• Innovated tax evasion countermeasures, recovering over €50 million.
• Pioneered State Electronic Tax Declaration checks, recognized as one of Brazil’s top taxation innovations in 2022.
• Developed models for predicting State Revenue from electronic invoices, aiding in government financial planning.
• Refined the methodology for calculating the State Tax Gap, focusing on high-evasion sectors from a macroeconomic perspective.
• Designed and implemented dashboards and reports to optimize tax auditing processes.
• Orchestrated strategic planning for State Tax Audits.
• Created and implemented models for swift detection of Shell companies, significantly reducing fraud detection time.

Headquater - Teresina - Brazil

• Tax Audit in the largest companies in the State.

Maintenance, Repair and Overhaul - São Carlos - Brazil

• Responsible for maintenance reports for Airbus 319/320/321 aircrafts.

Course in statistics and computer science at one of the most prestigious Italian universities

Graduation Grade: 109/110

Courses: Statistics, Econometrics, Algorithms, Machine Learning I and II, Mathematical modeling, Natural Language Program, Deep Learning.

Specialization course in Brazilian value-added tax legislation

Related courses: Tax Law, Constitutional Law, Value-Added Tax

5-year engineering course in Sao Carlos - Sao Paulo / Brazil

Specialization: Mechatronics and materials design.

Thesis Title: Study of biofuels and their viability in internal combustion engines.