Bruno (rosilva) Chicelli

2021 - 2023 (2y) ongoing
Masters of Science in Mechatronics Engineering
Escola Politécnica - USP
Dissertation on Reinforcement Learning for Traffic Control
=====
2020 - 2021 (2y) finished
Specialization in Artificial Intelligence
Escola Politécnica - USP
Dissertation on Computer Vision for Vehicle Classification and Counting
=======
2015 - 2019 (4y) finished
Bachelor's in Science and Technology
UFABC
=====
Online CoursesDeep Learning Specialization (4 courses, 16 weeks)
Deepmind.aiTensorFlor Developer Specialization (4 courses, 16 weeks)
Deepmind.aiMachine Learning with TensorFlow in GCP Specialization (4 courses, 16 weeks)
Deepmind.aiCoursera - 16 courses
Machine Learning, Andrew Ng and 15 others in the specializationsedX - 2 courses
Deep Learning foundations with Keras; Analyzing and Visualizing with PowerBIHarvardX - 3 courses
Data Science: Productivity Tools; Visualization; Machine LearningUdacity - 2 courses
Time Series Forecasting; Programming foundations with PythonOther courses (~20 more) are on my Linkedin page

Programming

Development

BI Tools and Cloud

Languages

Plotly Calendar Plot

• github repository
• Medium Article

Making it easier to visualize and costumize time relevant or time series data with plotly interaction.fig = calplot(dummy_df, x="ds", y="value", dark\_theme=True)
fig.show()

Todoist Analytics

• github repository
• MVP Link

If you use Todoist for daily, weekly, monthly or yearly planning you should try this tool.
This is how the current version of the tool looks like.The main goal for this tool is to help you keep track and create analytics based on your data from todoist.

NLP Job Classifier (based on linkedin data)

• github repository

A Job classification between Data Analyst, Data Scientist and Data Engineer web app made with Tensorflow, Selenium and deployed with Streamlit on Heroku. One simply inputs the job description in the input section and the model trained with ~3k job descriptions scraped with Selenium from jobs in the US.

HackerRank Scraper

• Github Repository

This repo is a scrapper to get my own solutions of HackerRank problems and to save them to my github repo.

SKLearn Playground

• github repository

This is a Playground for Scikit Learn, there's a live demo hosted on Heroku (link in the repo readme). My main objetive is to give the user an interface to tweek parameters and explore Scikit Learn models

XNatureV2 Classifier

• Github Repository

In this project I build an image classifier of x-ray images using numpy, keras and openCV
This is a diagram of the Neural Network using NNSVG

Machine Learning Animations

• Github Repository

Some ready to go animations for you to make your own Statistical, Mathematical or ML animations.
All gifs were made with TensorFlow, Matplotlib, sklearn and Celluloid
import DecisionTreeClassifierAnimation
anim = DecisionTreeClassifierAnimation()
k = anim.animate_training(x\_train, y\_train, x\_test, y\_test)
k.save("..\\animations\\decision\_trees.gif", writer='imagemagick')

Crossing Detection

Github Repository

This project was has the objective of detecting people faces crossing a line using an imaginary line across the image and using face landmark detection to detect when face keypoints cross the line.
The face and landmark detection was made using MTCNN.

Professional Experience

Staff Data Scientist - AdTech @ Globo
april/2023 - current

Recently promoted

Senior Data Scientist - AdTech @ Globo
july/2021 - march/2023

Skills: Git · Software Development · Computer Vision · Statistics · TensorFlow · PyTorch · Kanban · Natural Language Processing (NLP) · Amazon Web Services (AWS) · Google Cloud Platform (GCP) · Python · Machine Learning · SQL · Go
- Created an automated midroll placement AI with minimal UX impact for thousands of videos per day at Globo with event-based cloud tools and deep learning;
- Created a docs synchronization tool for several platforms at once using Go, Hugo, Confluence, Sharepoint and Tsuru;
- Created the Globo Tech Paper Club for discussions of scientific papers;
- Contributed in several data products for digital inventory optimization and forecasting, pricing optimization, sales forecasting and other projects.

Data Scientist - Advanced Analytics Team @ Boa Vista
august/2020 - june/2021

Skills: Git · Software Development · Statistics · Keras · Kanban · Google Cloud Platform (GCP) · Python · MLOps · Deep Learning · Machine Learning · SQL
- Developed a fast modelling tool to reduce Data Scientists overhead using LGBM, XGBoost and CatBoost;
- Created a model for propensity of credit card acceptance and activation;
- Created several improvements in the standard of modelling processes.Responsible for the development of innovations in MLOps processes - such as Deep Learning - and workflow automations, aiming to improve the impact, productivity and empower other data scientists in the Data Science Department

Jr Data Scientist @ ArcelorMittal
march/2019 - july/2020

Skills: Azure · Computer Vision · Python · Keras · R · Power BI · Selenium
- Saved ~1.3M BRL per year with automation and data science;
- Created a sales forecasting project for over 100 points of sale all over Brazil;
- Customer Churn modelling;
- Demand prediction for factories;
- Statistical modeling Python, R and using Machine Learning for forecasting and generating insights from analysis;
- Enrolled in projects envolving analytics in several areas.

Jr Data Analyst @ ArcelorMittal
july/2018 - february/2019

Skills: Azure · SQL · Python · R · Power BI · Selenium · VBA
- Created a PowerBI culture with 7k+ monthly views;
- Finding KPIs to provide the cleanest, simplest and best way to visualize data;
- ETL, automation and UX in PowerBI reports;
- Dealing directly with directors and head managers in sales, planning, logistics.