Social, Environmental and Climate Risk Analyst
Sicredi Internship
- Training and refinement of machine learning models and analytical forecasting models.
- Automation and analysis of routine processes.
- Data analysis and presentation.
Currently at Sicredi · Porto Alegre, Brazil
Davi Augusto
Data scientist & statistician in training.
I love turning real-world problems into questions that data can answer. I work end to end, from collecting and preparing data to modeling with machine learning and AI, always with statistical rigor and clear communication of results.
In [1]: LinearRegression().fit(X, y)
Click to add points · drag to move them
In [1]: davi.info()
About me
davi.info()
- location
- Porto Alegre, Brazil
- education
- Statistics @ UFRGS
- current
- Sicredi (intern)
- stack
- Python · R · SQL
- focus
- Data Science · ML · AI · ML Engineering
- languages
- Portuguese · English
I’m a Statistics student at UFRGS (Brazil), graduating in December 2026. I started in 2019 as a programming apprentice at Grupo RBS, already building reports and dashboards, and have been moving towards where code meets statistics ever since: analyses, KPIs, dashboards and predictive models, with stops at EVCOMX and Auroque Investimentos.
Today I’m at Sicredi, in Social, Environmental and Climate Risk, training and refining machine learning and forecasting models and automating analytical routines. For my thesis, I’m investigating what makes a game popular on Steam, with more than 170k apps collected through its API.
I care about reproducible analysis: code, text and results side by side, so anyone can check how I got to every number.
Skills & toolkit
Languages & tools
- Python
- R
- SQL
- SAS
- Databricks
- Git
- Jupyter
- Quarto
- R Markdown
Machine learning & AI
- scikit-learn
- TensorFlow
- Keras
- PySpark
- statsmodels
- spaCy
Modeling
- Linear & logistic regression
- GLMs (Poisson, negative binomial)
- Random Forest
- Gradient Boosting
- Neural networks
- Time series
- Clustering
- PCA
- NLP
Statistics
- Inference
- Hypothesis testing
- Bootstrap
- Sampling
- Multivariate analysis
- Spatial statistics
Solutions
- Predictive models
- Forecasting
- Dashboards
- KPIs
- Process automation
- API data collection
- Data apps
Data & visualization
- pandas
- NumPy
- Matplotlib
- Seaborn
- Plotly
- ggplot2
- Streamlit
In [2]: df_experience.head()
Experience
Four companies since 2019, from full-stack development to data science.
- Work
- Education
- today
Data Intelligence
Auroque Investimentos Internship
- Development and maintenance of databases, ensuring data integrity and management efficiency.
- Interactive dashboards built with Python and Streamlit.
- Custom KPIs for performance monitoring, supporting strategic decision-making.
- Data cleaning and filtering, improving overall data quality.
Data Science
EVCOMX Internship
- Data analysis in Python to identify trends and patterns, generating insights for the company.
- Machine learning models for prediction and scenario analysis.
- KPIs for performance monitoring, contributing to continuous process improvement.
- Statistical and descriptive analysis, including data cleaning and filtering.
Systems Programmer
Grupo RBS Apprentice
- Full-stack development focused on data analysis solutions in R and Python.
- Reports and data visualization dashboards using BI tools.
- Statistical techniques applied to support strategic decision-making.
In [3]: projects.nlargest(5, "impact")
Projects
From data collection to communicating results, some of what I've been building.
Undergraduate thesis · UFRGS
in progressPredicting game popularity on Steam
An end-to-end pipeline, from data collection to modeling, that predicts how many reviews a Steam game gets from its own characteristics, such as genre, price, platforms and description text.
- apps collected from the Steam Web API
- 171,769
- predicting whether a game gets reviews (gradient boosting)
- AUC 0.899
- Resilient collection with atomic checkpoints that resume where they stopped; 118k games in the curated dataset.
- Time-based train/test split and a hurdle model for the 20.6% of games with no reviews at all.
- Poisson vs. Negative Binomial GLMs (overdispersion, likelihood-ratio test). Next step, neural network modeling (RLMP).
More projects
Palmer Penguins: one-sample inference
Group project for Introduction to Inference (UFRGS) with descriptive analysis, confidence intervals, bootstrap and hypothesis tests, presented in Quarto.
Phone usage: two-sample inference
Comparing proportions between two populations with confidence intervals and hypothesis tests, also in R and Quarto.
NLP apps with Streamlit
Two apps to practice NLP and deployment. One runs statistical text analysis on PDF, DOCX and TXT files; the other generates automatic summaries.
More on GitHub
See all repositoriesIn [4]: history = davi.fit(courses)
Education
– (expected)
B.Sc. in Statistics
UFRGS · Federal University of Rio Grande do Sul
- Thesis: Predictive analysis of game popularity on Steam using neural networks, advised by Prof. João Henrique Ferreira Flores, PhD.
- Statistical consulting through NAE, the university’s statistical consulting unit, as part of the Laboratory 2 course.
- Courses with projects on GitHub: Inference, Linear Models, Multivariate Analysis, Sampling, Spatial Statistics and Computational Methods.
Courses & certificates
AI Engineer for Data Scientists Associate
Data Science, Machine Learning with Python
Data Scientist Training: The Complete Course
Statistics: Concepts and Representations
English
In [5]: blog.head()
Blog
Notes on statistics and data science, with code.
Posts are written in Portuguese.
Como Preparar uma Regressão Linear?
Construindo uma regressão linear simples à mão em Python (mínimos quadrados, R², RMSE e regra empírica) e conferindo os resultados com statsmodels e scikit-learn.
In [6]: davi.contact(subject="opportunity")
Let's talk
I'm open to conversations about data, statistics and data science opportunities. The fastest way to reach me is email or LinkedIn.