Computer scientist who have worked with a lot of different data positions such as data analysis, data science, ml engineering and even as a full stack developer. I am always learning and posting what I learned in Medium. I also worked on a platform to help the (non-governmental organizations) NGOs act at the Brazillian legislative process, on a really interesting computer vision problem using deep learning to LG, recommender system for a disruptive e-commerce start up, NLP system for Epson, a lot of cool stuff at Cloud Humans such as NLU chatbot, tickets classification models, audit system and LLMs products. Right now, I’m working at Stone. Skills: Python, R, Tensorflow, Keras, scikit-learn, pandas, NumPy, Data Science, NLP, LLMs, Airflow, Databricks, MLFlow, GCP, Azure, Dialogflow, OpenAI, Docker
Python
Kotlin
Javascript
Vue.js
Working creating bronze, silver and gold datasets.
Working creating dashboards using Looker Studio and Tableau
Worked on building the company's AI best practices and the data team's culture. Integrate machine learning models with the backend and front end of the company. Constructed a model for finding bad tickets that increased the audit process performance by 87%. Built chatbots that were responsible for answering 40% of the tickets Built from the ground a new product using LLM, that made the company double the size of clients in 2 months. Technologies used: python, databricks, mlflow, GCP, git, SQL, Airflow, LLM, OpeanAI, Kotlin, Docker, Spark.
LG Electronics
- Implemented the 3 most used Object Detection Algorithms, YOLO V3, SSD, and Faster RCNN using python, TensorFlow, Keras, NumPy, and other most used libs in the data science field.
- Implemented the unprecedented D-RISE technique for Object Detection model explainability.
- Improved the SSD model to get a mAP of over 0.90.
E-commerce startup
- Implemented several recommender system algorithms using Python, which increased sales by 105%.
- Designed experiments such as the A/B test.
Epson
- Worked with NLP (Natural Language Processing) to build a DocumentIA approach, that increased in 64% the process of filling forms
Skills: Machine Learning · Python (Programming Language) · Recommender Systems · Data Science · Nlp · Computer Vision · SQL
I worked with a team of 8 since the beginning on the development of a system to help NGOs and the population with the transparency and efficiency of the Brazilian legislative process. It was a really challenging job because I had to work at all parts of the development of the system from the DevOps and backend part until the front end and team management.
I worked gathering, cleaning, organizing, and extracting relevant pieces of information from the data mostly using R and python for the machine learning part. I worked too with python (Django) on the back and with Vue.js on the front, using D3 and vega-lite to build amazing visualizations. In addition to that, we worked using SCRUM, I and a colleague were responsible for managing the team, we always tried to make sure that the team was synchronized, each team member was working with the field that the member was more passionate about while making sure that everyone was happy.
On this adventure, I had to be a flexible and creative person because I had to deal with different parts of a business, from hiring accountants, designers, and developers to team management and software engineering.
This opened my mind to start-ups' culture of problem-solving and innovation.
Manage a Team of 2 people using SCRUM.
- Built an IOS app having 12 screens using Swift
I worked a little with angular and javascript on the front end of a website for improving the public transportation of Campina Grande, I worked with python and flask on the backend and a little with R to make some analysis of the trips.
Worked too on an app for the users of the public transportation of Campina Grande, coded in Java using the Android SDK.
Site link: www.facebook.com/analytics.ufcg/
