li { border: 1px solid grey; }
Hi! Hello and welcome to my website! I am Hugo!
You can find some of my blog posts here. In addition, you can check out or contribute to any of my projects by clicking here.
I am a computer scientist currently working in computer vision systems. I have 5+ years of work experience as a machine learning scientist, data scientist, and software engineer. In addition, I frequently work in the field of game development.
I successfully achieved my Master's degree in Electrical and Computer Engineering, Machine Learning from Université Laval (Quebec) and my Bachelor's degree in Computer Science from the Federal University of Rio de Janeiro (Brazil).
My passion for building intelligent machines tells me that reasoning is an effective means for minimizing most human errors. In every project, I remain committed to understanding the data and using it to solve difficult tasks. I'm always looking for ways to grow, not just in my career, but also in philosophical discussions and curiosity-driven learning. My ultimate goal is to create a world in which technology enhances human capabilities, and I do believe that logic with data, is the key to achieving that.
In my spare time, I enjoy having philosophical discussions, listening to video game music, and learning meaningless things like typing faster.
July 2022 - Currently
- Development of 2D and 3D AI tools in a vision software forinline measurement and inspection solutions.
- Development of smart camera projects with 2D/3D tools and calibration methods.
- Designed and deployed a real-time Anomaly Detection pipeline, leveraging deep learning to streamline workflow for labeling, training, and testing models
- Ported and optimized vision tools pipeline to a vision software of 3D smart sensors, using C, C++, and Python.
- Developed GoAIK module, enabling integration between with GoPxL and Cognex, a leading machine vision software, improving data accuracy and processing time.
- Technologies: C, C++, Python, Deep Learning, Anomaly Detection, Computer Vision, Machine Learning.
Oct 2021 - July 2022
- Implemented a pick and place robotic solution using Universal Robots and Realsense/Kinect/Ensenso cameras.
- Implemented a computer vision pipeline (data labeling for 3D segmentation, training and evaluating model) for point cloud classification using PointCNN.
- Implemented a production-level QT desktop application for a robotic tool.
- Technologies: Reinforcement learning, Open3D, Pytorch, Universal Robots, Computer Vision.
Apr 2021 - Oct 2021
- Implemented a hand tracking system to work in low-light conditions using numerous realsense cameras.
- Implemented a pose estimation system and OSC server in Unity Engine.
- Technologies: Machine learning, Computer Vision, Unity Engine, RealSense, OSC, Python, and C#.
Nov 2017 - May 2018
- Implemented a manageable infrastructure forimage processing in cloud servers reducing process time by 30%.
- Automatized time-consuming tasks (10TB training data for annotation) using SOTA deep learning methods.
- Presented key findings for end customers to the team leader and wrote an executive report.
- Technologies: Deep Learning, Computer Vision, AWS, Docker, and Python.
Global Research
Nov 2016 - Nov 2017
- Deployed 100+ wearable applications that gather training data and send to docker containers.
- Processed 1TB of previously inaccessible data for a Human Activity Recognition system, improving prediction accuracy by 50% and speed by 400% forreal-time client impact tracking.
- Technologies: Deep Learning, Time Series, Docker, Python, C++, and Javascript.
December 2020
Laval University
- Thesis: Meta Learning Population-Based Algorithms in Black-box Optimization
- Fellowships:
- Relevant Coursework:
- Deep Learning
- Deep Reinforcement Learning
- Statistical Learning Theory
December 2017
Federal University of Rio de Janeiro
- Thesis: Towards Deep Q-Caching
- Fellowships:
- National Council of Scientific Researches Scholarship
- Institutional Scientific Initiation Scholarship
- Teach Assistant (TA):
- Data Structures and Algorithms
- Reinforcement Learning
2021
Meta Learning Black-Box Population-Based Optimizers
Hugo Siqueira Gomes , Benjamin Léger, Christian Gagné
The no free lunch theorem states that no model is better suited to every problem. A question that arises from this is how to design methods that propose optimizers tailored to specific problems achieving state-of-the-art performance. This paper addresses this issue by proposing the use of meta-learning ...
2019
Unsupervised Temperature Scaling: Post-Processing Unsupervised Calibration of Deep Models Decisions
Azadeh Sadat Mozafari, Hugo Siqueira Gomes , Wilson Leão, Christian Gagné
The great performances of deep learning are undeniable, with impressive results over a wide range of tasks. However, the output confidence of these models is usually not well-calibrated, which can be an issue for applications where confidence ...