Author Image

Hi, I am Marcos

Marcos

Master's Student at INTRIG group - Unicamp

I am passionate about machine learning and I’m always doing projects and researching. The possibility of having an idea that can help someone is what motivates me to work and learn about data science every day.

In my master’s degree I study ML-based unsupervised anomaly detection in time-series applied to optical fibers.

I’m also a Trainee at a startup where I develop computer vision solutions with YOLO, perform Front-end with Angular, Back-end tasks with C# and data analysis with SQL and Power BI.

Python
Machine Learning
Research
Data Analysis
Team Work
Computer Vision

Recent Projects

Card image cap
ML anomaly detection in time-series

Development Start: 08/2022. End: In Progress… Project description I’m doing a master’s degree at the Intrig electrical engineering group at Unicamp. My goal is to create a digital twin network from the fiber optic networks of the company PadTec from Campinas. I participated in the application of a gradient descent algorithm on the digital twin. And at the moment I am collecting practical data from the network in order to model the network with real data.

May 17, 2023 Read
Card image cap
MBA Big Data and Business Analytics Projects

Development Start: 04/2022. End: 06/2023. Project description (MBA Page) Big Data is the term used to describe the vast volume of data that impacts business on a day-to-day basis. The Executive MBA in Business Analytics and Big Data enabled us to analyze business problems and use analytical techniques in this current scenario characterized by complexity, diversity and high volume of digital data. During the course we learned about: Analytical ability: To manage and conduct projects involving structured and unstructured databases; Business management: Ability to analyze and identify problems and generate solutions to business problems, based on the practical application of methods; Analytical Problem-Solving: Possessing extensive expertise in data modeling, quantitative analysis, and problem identification/resolution, enabling effective decision-making and driving successful outcomes.

May 10, 2023 Read
Card image cap
Machine Learning Guide

Development Start: 10/2022. End: 06/2023. Project description (Link to guide) I made a machine learning guide that includes all stages of its development, main techniques and algorithms developed to date, as well as websites, codes and auxiliary tutorials. It has more than 300 pages, and can help you discover new tools and algorithms that you may never heard of before. Chapters Include: Ethics Concepts Pre-Processing Feature Selection Regression Classification Clusters Recommendation Time-Series Neural Networks Reinforcement Learning Ensemble Methods Explainable Models (XAI) AutoML Cost-Sensitive Learning Dimension Reduction Anomaly Detection Model Tuning Evaluation metrics Statistics Model Compression Python and R tools Storytelling with Data AI tools Tutorials, Courses, Code Examples, Books And more…

May 10, 2023 Read

Experiences

1
Trainee
Mottu

November 2023 - Present, São Paulo, Brazil

Prevention and Loss Trainee to improve the management of the company’s motorcycle fleet

Responsibilities:
  • Data analysis with Excel, Google Cloud Platform (GCP), SQL and development of dashboards with Power BI.
  • Development of a computer vision project for license plate recognition with YOLOv8 and OCR using Python, Docker and Big Query to avoid losses and better manage a fleet of 70 thousand motorcycles.
  • Led the planning and execution of an application for monitoring the entry and exit of motorcycles using UX design, Figma and Flutter to be implemented in more than 50 branches.
  • Development of webpages with Angular and API creation with C#.
AI Researcher
Samsung Electronics

Jan 2023 - Mar 2023, Campinas, Brazil

AI research using bayesian neural networks

Responsibilities:
  • Study the explainability of the existing models.
  • Mitigate uncertainties and generate new insights for the project.
  • Develop a Bayesian Neural Network applicable to any model using pytorch library.
2
3
Sales Business Analytics Intern
Siemens Energy

Jan 2021 - Dec 2021, Jundiaí, Brazil

Internship to help and monitore the sales team to achieve their goals

Responsibilities:
  • Registration and monitoring of opportunities in Salesforce.
  • Development of Power BI dashboards with our sales goals and KPIs.
  • Search for new opportunities in the market.
Equipe Phoenix de Robótica da Unicamp

Mar 2017 - Jun 2019, Campinas, Brazil

Extracurricular robotics project for student competitions

Marketing coordinator

Aug 2017 - Jun 2019

  • Social Media Posts, creation of visual art for flyers.
  • Creation of the website.
  • Organization of the selection process.
Electric department member

Mar 2017 - Jun 2019

  • Member of the Trekking project, autonomous robot, national category champion in 2017 at the Winter Challenge.
  • Featherweight(30lb) category ramp combat robot project member.
  • Projects with Altium to make circuit boards.
  • Projects with Arduino to work with sensors.
4

Education
UNICAMP
2022-2023
Master's in Computer Science
CGPA: 4 out of 4
Publications
FGV - Faculdade Getúlio Vargas
2022-2023
MBA in Big Data & Business Analytics
CGPA: 9 out of 10
UNICAMP
2017-2021
B.Sc. in Electrical Engineering
CGPA: 7.5 out of 10
Extracurricular Activities
  • Phoenix robotics team
  • E-racing electric cars
Instituto Superior Técnico
2019-2020
Student Exchange Program
Taken Courses
  • Image video processing
  • Electrical Machines
  • Machine Learning
  • Photovoltaic Systems
Liceu de Artes e Ofícios de São Paulo
2012-2014
Electronics technician
CGPA: 7.5 out of 10

Skills

Achievements

Winner EY Business Challenge 2020

Certificates

Google Cloud Courses Badges
Google

Includes Generative AI Fundamentals, Big Data and Machine Learning Fundamentals and much more.

View Certificate
Google UX Design
Google

Start the UX Design Process and Build Dynamic User Interfaces (UI) for Websites.

View Certificate
Microsoft Power BI for Data Science, Version 2.0
Data Science Academy Apr 2021

Online course with a workload of 72 hours to develop dashboards with Power BI.

View Certificate
DP-100 A-Z Machine Learning using Azure Machine Learning
Udemy Mai 2022

Online course with a workload of 26.5 hours.

View Certificate
Machine Learning A-Z™ Hands-On Python & R in Data Science
Udemy Apr 2021

Online course with a workload of 44 hours.

View Certificate
Deep Learning A-Z™ Hands-On Artificial Neural Networks
Udemy Sep 2020

Online course with a workload of 22 hours.

View Certificate
Markov Matrices
MOOC - Tecnico Lisboa Apr 2020

Online course with a workload of 8 hours.

View Certificate
Drones Simulation and Control
MOOC - Tecnico Lisboa Apr 2020

Online course with a workload of 8 hours.

View Certificate

Overview

Davai
Founder and Developer Jan 2021 - Mar 2022

Music social network.

professional app flutter
Details
Football Simulator
Owner Jun 2020 - Present

A Football Simulator App developed with Flutter.

hobby app football
Community Supportive App
Owner Dez 2020 - Dez 2020

“Solidário” app a cashback system to help the community with donnations.

academic app
Football standings prediction using time-series
Owner May 2024

Prediction of a football league final season classification based on previous performances.

hobby machine-learning football
ML Guide
Owner Dec 2022 - Present

A Machine Learning containing a resume of all the most used techniques in the field.

hobby machine-learning neural-networks
YOLO - Pedestrian and license plate detection
Author Jun 2020 - Dec 2020

Image detection with YOLO framework.

academic machine-learning flutter
Details

Publications

Unsupervised Machine Learning for Anomaly Localization in Optical Amplifiers using Real Network Telemetry Data
2024 European Conference on Optical Communication (ECOC) 28 April 2024
Marcos P. A. Dal Maso Kayol S. Mayer Andres F. Escallón-Portilla Darli A. A. Mello Dalton S. Arantes Christian E. Rothenberg

We propose an unsupervised ML approach to localize optical amplifier anomalies using input-output power time series. Tested on real optical network telemetry data, our method successfully identified a faulty amplifier.

Machine Learning time-series Optical Communications anomaly-detection
Details
Network-Wide QoT Estimation Using SGD with Gradient Transfer Between Wavelengths.
2023 European Conference on Optical Communication (ECOC) 5 May 2023
Kayol S. Mayer Rossano P. Pinto Marcos P. A. Dal Maso Humberto V. Q. Melo Christian E. Rothenberg Darli A. A. Mello

We propose an SGD-based QoT estimation technique that operates on a network-wide scale by transferring gradients among neighboring wavelengths. Simulation results indicate effective and low-complexity QoT estimation using only transponder SNR telemetry.

Machine Learning Digital Twins Optical Communications
Details