CONSULTING · BACKEND
APIs and enterprise integrations
I built REST APIs that connect internal platforms to ERP and CRM systems and automate steps in operational workflows. In production, I monitored the integrations and used logs to investigate incidents.
Technical contribution
Alongside the APIs, I built internal tools to automate repetitive operational tasks that previously required manual work.
CONSULTING · DATA ENGINEERING
Profit-sharing eligibility pipeline
I built a Databricks pipeline that combines HR and leave records to calculate days worked and produce an eligibility dataset for HR to review by employee and six-month cycle.
Technical contribution
I split leave periods across cycle boundaries and consolidated employment records for the same employee. When validating the output, I checked cases involving changes in employment records and end-of-cycle departures to catch double counting and records carried into the following cycle.
CONSULTING · MODEL VALIDATION
Transaction model validation
I worked on validating transaction models for individual and business banking customers.
Technical contribution
I reviewed input data quality, technical documentation, and whether model runs could be reproduced.
CONSULTING · BUSINESS INTELLIGENCE
Semantic modeling for Power BI reporting
I worked on organizing Databricks data into a star schema for analysis in Power BI.
Technical contribution
I modeled fact tables and shared dimensions so reports could apply consistent filters across datasets.
SOFTWARE · FULL STACK
Time-tracking system
I developed an internal system for recording and tracking work hours.
Technical contribution
I implemented the backend logic for time entries and the interface used to log and review hours.
AUTOMATION · FINANCIAL OPERATIONS
Financial process automation
I developed an automation for a financial workflow involving data from CRM and ERP systems.
Technical contribution
I used RPA to connect steps in the workflow across the two systems and reduce manual work.
CAPSTONE · PUBLISHED PAPER
Fire and smoke detection in images
I co-authored a study comparing three YOLO versions for fire and smoke detection. We used 870 images from two public datasets, split into training, validation, and test sets.
In our experiment, YOLOv8s detected more fire and smoke instances, while YOLO11s produced fewer false positives.
Revista Contemporânea · 2024
Read the published paper →