João Braz Simões
Machine Learning Engineer/Researcher at CISUC - Centre for Informatics and Systems of the University of Coimbra- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
Portuguese Native or bilingual proficiency
-
English Full professional proficiency
-
Italian Elementary proficiency
Topline Score
Bio
Credentials
-
Convolutional Neural Networks
CourseraNov, 2020- Oct, 2024 -
Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
CourseraOct, 2020- Oct, 2024 -
Structuring Machine Learning Projects
CourseraOct, 2020- Oct, 2024 -
Neural Networks and Deep Learning
CourseraSep, 2020- Oct, 2024
Experience
-
CISUC - Centre for Informatics and Systems of the University of Coimbra
-
Portugal
-
IT Services and IT Consulting
-
1 - 100 Employee
-
Machine Learning Engineer/Researcher
-
Sep 2020 - Present
I have been building ML models in several contexts. I'm currently developing deep generative models that produce synthetic spatial graph structures, which simulate complex systems in various fields. I am proposing an encoder-decoder model (also called sequence-to-sequence model) to generate spatial structures. I am soon releasing a new public PyTorch framework that will allow for the generation of synthetic spatial graph structures from existing data. I have already applied this framework to generate vascular networks, converting 3D imaging data to graphs and using it as the training set. In a similar context, I have previously proposed a Deep Reinforcement Learning (Q-Learning) model to generate spatial graphs, in particular vascular networks. Skills and Technological stack: - Python - PyTorch - Numpy - scikit-learn - Integration with experiment management tools (neptune.ai and wandb) - Docker for replicating the working environment and quickly train models in external clusters - Speed-up ML models through GPU-acceleration - Dive deep into new fields (e.g. learn the basics of biology to model vascular networks) - Apply Reinforcement Learning techniques (e.g. Q-Learning, PPO) Show less
-
-
-
CRON.
-
Portugal
-
IT Services and IT Consulting
-
1 - 100 Employee
-
Co-Head of Software
-
May 2019 - Oct 2021
I was (co-)responsible for managing the software department at Cron.Studio, which encompassed coordination, client management, software development, project estimation and strategic planning for the software department. Cron.Studio is a startup studio that develops technological solutions from the very beginning to the first MVP. As co-head of software, I was responsible for successfully delivering technological products for startups operating in different fields: smart-lockers, water management, fitness and health and disruptive research. In my technical duties, I was involved in the process of defining the technological stack and designing the architecture of solutions composed by several connected components (backend, mobile/web clients and external bluetooth devices), giving also my contribution in the implementation phase. In the context of Machine Learning solutions, I was the tech lead in a project with a startup operating in the field of sustainable water management. Their solution consists in a web-based platform to monitor water networks, automatically detecting leakages. The high amount of data from their built-in sensors was used to build a model capable of detecting water leakages. During my journey in Cron.Studio, I worked with the following technological stack: - Python - Backend with Django - Firmware with Python - Redis - RabbitMQ - scikit-learn - xgboost - Machine learning with time-series for failure detection in water networks - React and React.Native Show less
-
-
-
Sentilant
-
Portugal
-
Software Development
-
1 - 100 Employee
-
Software Engineer
-
2017 - Apr 2019
Backend development with Python, using technologies such as Django, Postgres, MongoDB, Redis, RabbitMQ, Ansible, and Docker. Additionally, I was responsible for developing, improving, and customizing a vehicle route optimization system built in Java. This involved incorporating a machine learning model to predict heavy vehicle accidents. The model was trained with historical data on past accidents, weather conditions, driver history, etc. Backend development with Python, using technologies such as Django, Postgres, MongoDB, Redis, RabbitMQ, Ansible, and Docker. Additionally, I was responsible for developing, improving, and customizing a vehicle route optimization system built in Java. This involved incorporating a machine learning model to predict heavy vehicle accidents. The model was trained with historical data on past accidents, weather conditions, driver history, etc.
-
-
-
University of Coimbra
-
Portugal
-
Higher Education
-
500 - 600 Employee
-
Grant Researcher
-
Sep 2016 - 2017
I was a researcher in Evolutionary and Complex Systems Group (part of the CISUC) and Center for Computational Physics, working on the field of computational biology, exploring and building a computational model to simulate the angiogenesis process (the process by which new blood vessels grow from existing ones), in an attempt to understand the link between this process and the progression of type-2 diabetes.
-
-
Grant Researcher
-
Jun 2015 - Aug 2016
Research grant for the project "Multi-Agent Simulator of Aircraft Evacuation"
-
-
Researcher
-
Oct 2014 - Jun 2015
Open-ended Evolution of Pheromone-based Stigmergic Communication
-
-
-
Dognædis
-
Portugal
-
Computer and Network Security
-
1 - 100 Employee
-
Summer Internship
-
Aug 2013 - Sep 2013
Web development of Dognaedis platform. Web development of Dognaedis platform.
-
-
Education
-
2014 - 2016
Universidade de Coimbra
Master’s Degree -
2011 - 2014
Universidade de Coimbra
Bachelor’s Degree, Computer Science
Community
