My Spider Chart

#

My Data Toolkit

#

Projects Beyond the 9-5

#

• From 2017 to 2021, I taught basic statistics to students of UFPR and PUC, mostly.
• I assisted researchers with statistical modeling for their master’s and PhD projects. Some of the most interesting ones included analyzing variables that impact fish growth in controlled environments and comparing the shelf life of donated blood between overweight and non-overweight donors.
• I also participated in the early stages of a major project in Paraná to measure the spread of dengue.
• Some of my graduation work you can find on this repository /FACULDADE: Risk Theory, Industrial Process Control, Extension of Linear Models, Evaluation Theories, Stochastic Processes, Computacional Methods in Statistical Inference, Text Mining
• During my undergraduate studies, I served as a teaching assistant for two courses: Descriptive and Inferential Statistics, and Linear Regression.

My Career Path

#

1. AGRINVEST | CONSULTANT Data Enginner

   2024/MAY-PRESENT

   Curitiba, Brazil

   Company Website

   If you read from bottom to top, you’ll notice quite a bit of R Shiny throughout my career. It turns out I’ve gotten pretty good at it, and nowadays, it's one of my core skills. I really enjoy building Shiny applications and working with R. In my current role, I’ve had the opportunity to rewrite older Shiny applications, making them more readable and optimized, while also building and maintaining an API.

2. CAPTAL | Data Scientist/Data Enginner

   2021/OCT-PRESENT

   São Paulo, Brazil

   Company Website

   Reference: Victor Boso

   Here I am—Captal came right after Kzas, and I continued to build on my real estate expertise. We leveraged some of the technologies we had previously developed—like web scraping, preprocessing, and Automated Valuation Models (AVM)—but now with a focus on identifying prime investment opportunities in the real estate sector.

   Later on, I spearheaded the automation of the feasibility models we had been running in Excel, transforming them into a fully-fledged software solution (again, using R Shiny). This system went beyond just the models, becoming an entire pipeline for originating real estate assets. It allowed us to assess feasibility, conduct real estate analyses (like checking square meter prices, comparing similar projects, and analyzing sales-to-inventory ratios), and even incorporate a bit of urban accessibility analysis, an idea I picked up from reading this [book](https://ipeagit.github.io/intro_access_book/).

   In addition to that, I supported the general infrastructure, working with tools like Airflow and handling exploratory data analyses (EDAs), among other responsibilities.

3. Gaia | CONSULTANT

   2022/MAY-2023-JUL

   Cidade del Mexico, Mexico

   Company Website

   Reference: Brandon Proctor

   This job came as a bit of a surprise. The marketing director from my time at MadeiraMadeira reached out and asked me to do a similar role, but with some new challenges. I built an architecture on AWS to crawl e-commerce websites, this time focusing on Mexico, and developed a forecast for the entire catalog using a top-down time series approach. I also provided support by creating analytical reports. It was a great experience overall, and I even picked up a bit of Spanish along the way!
4. MadeiraMadeira | CONSULTANT

   2021/APR-2022-JAN

   Curitiba, Brazil

   As I mentioned earlier, I continued supporting my previous company even after moving on. I primarily helped with A/B testing and assisted with maintaining a software tool I had built to manage SEO. The tool allowed us to customize SEO strategies for different product categories where the generic approach wasn't performing well.

5. Kzas | Data Scientist

   2021/APR-2021-SEP

   São Paulo, Brazil

   At one point in my career, I had the opportunity to lead a very exciting project from start to finish: building an Automated Evaluation Model for Real Estate. It was originally planned to take a year, but we managed to complete it in just six months.

   It was a challenging time for me because I didn’t want to leave my previous company, and I had started this new role as a contractor. So, while working on this project, I continued to provide support to my previous company, no longer as an employee but as a contractor.

   As for the project itself, we developed numerous web crawlers for real estate websites using Python (Scrapy). We then preprocessed the data to ensure its quality and designed a solid architecture to automate the entire process.

   After that, we spent about a month and a half developing the best model to price properties in São Paulo, achieving a 10% error rate. We also built an API to deliver the results and an interactive interface (using Shiny again) so users could easily test the model. Once my part was done, I handed it over.

6. MadeiraMadeira | Statistical Analyst

   2019/NOV-2021-APR

   Curitiba, Brazil

   Reference: Brandon Proctor

   MadeiraMadeira was one of the best moments in my career, as it was my first real job (my previous experience was an internship). I can confidently say that this was my "laboratory," where I applied everything I learned in college. But more than that, it significantly shaped my skills, particularly in data engineering, which isn’t typically covered in a statistics degree. It was here that I first encountered AWS, Docker, Hadoop, and Spark. I built dozens of automated processes to calculate marketing KPIs—what I later realized were essentially ETL pipelines.

   I was involved in several projects, but one of the most significant was developing sales forecasts for the entire product catalog. This required frequent communication with end users, particularly when they raised concerns about why some products had similar sales forecasts (e.g., "Why are low-selling products predicted to perform the same?"). This experience taught me valuable lessons in both diplomacy and understanding the users' perspectives.

   Some of the major projects I worked on included optimizing sales across various marketplaces and building a dashboard with R Shiny that visualized the company’s logistics process. MadeiraMadeira is an e-commerce company, similar to Amazon, and with this dashboard, they could track delivery times across different cities throughout Brazil. I also built web crawlers for major e-commerce platforms to help improve business strategies.

   However, the two projects I enjoyed the most were the search algorithm optimization for product placement on the website and a text mining project analyzing customer reviews.

   For the text mining project, I identified the main issues customers were reporting in their reviews and relayed this feedback to sellers. By finding patterns across multiple products, we created a list of items to either negotiate better prices or, in some cases, remove from the catalog entirely.

   The search algorithm project was a personal favorite due to its complexity. When I first joined the project, the SEO process was incredibly complicated, relying on dozens of variables and a genetic algorithm to optimize business rules. I simplified it by implementing a weighted average based on just four key variables. This adjustment streamlined the entire process and led to the best possible product display on the website. Despite its simplicity, this change had a significant impact, boosting the conversion rate from 0.00003 to 0.011 on average. It was a real game-changer.

7. Bradesco Bank | Statistical Intern

   2018/MAR-2019-JUL

   Curitiba, Brazil

   This was my first experience as a statistician and data scientist. Working at the bank gave me invaluable insights into corporate processes and the inner workings of large organizations. I was amazed by how various statistical teams collaborated to create tangible value. I contributed to several stages of this process, from database extraction and normalization to enhancing the final reports and tracking model performance. Our main goal was to develop models that helped "recover" funds from clients who defaulted on their loans.

   Two key moments stand out from my time at the bank. The first was a friendly competition among the interns to suggest improvements for the bank’s strategies or environment. Honestly, I can’t recall exactly what we proposed, and here’s the plot twist—no one ended up winning!

   The second memorable experience was a report I independently created, analyzing the economic cycle of a Brazilian government-controlled index. I estimated the average duration of economic booms following a presidential election. This analysis became part of the bank's strategy to anticipate how credit would be affected during a time of significant political upheaval.