Automated phase-type distribution fitting via expectation maximization
Journal of Reliable Intelligent Environments
Recife, Brazil
Marco Mialaret Júnior
Senior Data Architect & Professor — I design data pipelines and agentic AI systems, and hold a PhD in Computer Science researching how phase-type distributions model system reliability.
Data architecture
Ingestion and integration pipelines across structured and unstructured sources, and data-driven system design.
Agentic AI systems
LLM-based agents and services — including an MCP server in FastAPI for orchestrating intelligent workflows.
Reliability & statistics
Modeling system failure times with phase-type distributions, fitted via Expectation-Maximization.
Teaching
Data Engineering, Statistical Inference and Data Analysis — teaching since 2011, from calculus to Big Data.
About
From algebraic geometry to data architecture
I hold a degree in Mathematics Education from UFRPE (2008) and a Master's in Mathematics (2011), with research in algebraic geometry. Around 2018 my interests shifted toward computational mathematics, and in 2023 I completed a technologist degree in Database Management at SENAC — the step that moved me fully into data.
In 2026 I completed my PhD in Computer Science at CIn-UFPE, researching statistical models for system failure times using phase-type distributions, estimated with Expectation-Maximization algorithms. Today I work as a Senior Data Architect at TJPE (via FSBR), building ingestion pipelines and agentic AI solutions, and I teach Data Engineering, Statistical Inference and Data Analysis at Cesar School.
Languages & data
Python
Java
SQL
PySpark
Analytics & AI
Machine Learning
LLM
Power BI
Cloud & tools
Amazon EC2
Amazon RDS
Git
Resume
Experience & education
Experience
-
Senior Data Architect
FSBR — Fábrica de Software do Brasil · allocated at TJPEIngestion and integration pipelines for structured and unstructured data, and data-driven architecture design. Building agentic AI solutions, including an MCP service in FastAPI to orchestrate intelligent workflows and automate processing.
-
Qualification Consultant — Professor
Cesar SchoolTeaching Data Engineering, Statistical Inference and Data Analysis.
-
Development Project Manager
Faculdade SENACLed "Hub do Comércio," a digital platform for Fecomércio-PE built on a data lakehouse architecture. Consolidated data from IBGE, Banco Central, Receita Federal and primary sources into interactive dashboards, automated reports and financial management tools.
-
Technical Instructor II
TrybeMentored students in Python and Java through practical projects — system maintenance, testing, code review and course material development.
-
Infrastructure Support
TecnosolveEnd-user technical support; diagnosed hardware, software, network and OS issues, and documented resolutions.
-
Assistant Professor
UFRPE · UNINASSAU · UNIFAVIP · SENACTaught foundational engineering courses — Calculus, Linear Algebra, Analytic Geometry, Probability & Statistics, Differential Equations, Numerical Methods — as well as Big Data, Data Science, Relational Databases and Project Management.
Education
-
PhD in Computer Science
UFPE (CIn)Research in reliability engineering, modeling system reliability with phase-type distributions. Applied Expectation-Maximization (EM) algorithms to estimate distribution parameters and contributed to their implementation in the research group's software.
-
Database Technologist
SENAC São PauloDatabase management fundamentals — SQL and NoSQL, data modeling, querying and data manipulation.
-
Master's in Mathematics
Universidade Federal da ParaíbaResearch in algebraic geometry; dissertation on static curves in the complex projective plane.
-
Degree in Mathematics Education
Universidade Federal Rural de PernambucoResearch on the local-global principle via the Hasse–Minkowski theorem, relating rational solutions of quadratic forms to their local completions.
Certificates
Neural Networks and Deep Learning
Improving Deep Neural Networks (Hyperparameter Tuning & Optimization)
Data Science: Solving Real Problems with Python & Pandas
Data Science: Data Visualization with Python
Support Vector Machine Classification in Python
Predict Future Product Prices Using Facebook Prophet
Foundations: Data, Data, Everywhere
Ask Questions to Make Data-Driven Decisions
Prepare Data for Exploration
Introduction to Git and GitHub
Publications
Research
Peer-reviewed work on reliability modeling, fog computing and applied machine learning.
Automated Reliability Analysis in Fog Computing Environment
Workshop on Resilience Engineering in Computing Systems (RECS), SBC
Building energy consumption models based on smartphone user usage patterns
Knowledge-Based Systems
A methodology for selecting hardware performance counters for supporting non-intrusive diagnostic of flood DDoS attacks on web servers
Computers & Security
Projects
Selected work
A sample from github.com/MatmJr — data pipelines, agentic AI and teaching materials.
AirportsProject
An end-to-end PySpark pipeline, containerized with Docker Compose, for processing airport data at scale.
LLMCreditCardBot
A natural-language-to-SQL chatbot powered by GPT-4o-mini, letting users query a delinquency and demographic database through a Streamlit interface.
ETL_IBGE_BCB
An ETL pipeline that extracts Brazilian Central Bank data, transforms it with Pandas, loads it into SQLite/CSV, and visualizes it through a Dash + Plotly dashboard.
End2EndTitanic
A Flask web app that predicts Titanic passenger survival probability using a Random Forest model.
CursoBigData
Lecture slides, notebooks and code used to teach Big Data — the most-starred repository on my GitHub.
Contact
Let's talk
Open to conversations about data architecture, agentic AI, reliability research or teaching.