Rafael Henrique Tibães, Developer in Curitiba, Brazil
Rafael is available for hire
Hire Rafael

Rafael Henrique Tibães

Verified Expert  in Engineering

Computer Vision Developer

Location
Curitiba, Brazil
Toptal Member Since
June 20, 2020

Rafael is a talented data scientist with expertise in credit analysis, insurance, and pricing. He has deployed AI models with real business and community impact. He developed algorithms for fingerprint matching under robust image scale variance caused by the individual aging, as part of the solution to help find missing children. A computer vision engineer with industry experience in biometrics, Rafael has solved complex face and fingerprint detection challenges, improving security and profits.

Computer VisionArtificial Intelligence (AI)Machine LearningOpenCVDeep LearningLinuxC++PythonTensorFlowScikit-learnPandasAI DesignData AnalysisData AnalyticsGitBiometricsDataOpsDlib

Portfolio

Logitech
Machine Learning, C++, Python, TensorFlow, PyTorch, Rust, Flutter
Junto Seguros
SciPy, Jira, DevOps, DataOps, ETL, Seaborn, Plotly, Jupyter Notebook, Jupyter...
Akiyama
DevOps, Augmentor, Dlib, CMake, AI Design, Canon...

Experience

Computer Vision - 8 yearsArtificial Intelligence (AI) - 8 yearsImage Processing - 8 yearsOpenCV - 8 yearsMachine Learning - 5 yearsPython - 4 yearsData Science - 3 yearsDeep Learning - 3 years

Availability

Part-time

Preferred Environment

Amazon Web Services (AWS), Scikit-learn, Git, TensorFlow, Pandas, Python, C++, Linux, Deep Learning, Machine Learning, OpenCV, Google Cloud Platform (GCP), Swift, Docker, MacOS

The most amazing...

...and complex solution |'ve developed was risk analysis for an insurance company using a set of machine learning models.

Work Experience

Machine Learning Specialist

2021 - PRESENT
Logitech
  • Designed and developed real-time data pipelines for a wide range of sources, like image, audio, and time-series.
  • Deployed embedded machine learning models for computer vision and audio.
  • Fed vision and audio models with real-time data from webcams and microphones.
Technologies: Machine Learning, C++, Python, TensorFlow, PyTorch, Rust, Flutter

Senior Data Scientist

2018 - 2021
Junto Seguros
  • Developed a robust solution composed by machine learning models to increase profits by accepting more clients and reducing the claim rate.
  • Performed data ingestion, cleaning, and balancing using SQL, Pandas, and Airflow.
  • Explored a range of algorithms and modeling strategies, such as tree-based models, clustering, and deep neural networks, including Feedforward, Autoencoders, and GAN architectures.
  • Ensured interpretability using SHAP values for features' importance and custom metrics for profitability and estimated claims.
Technologies: SciPy, Jira, DevOps, DataOps, ETL, Seaborn, Plotly, Jupyter Notebook, Jupyter, Data Analytics, Data Analysis, AI Design, Microservices, REST APIs, Serverless, AWS Lambda, Amazon Web Services (AWS), Pricing, Credit Risk, Insurance, Scikit-learn, Git, Scrum, TensorFlow, Pandas, Data Science, Python, Linux, Deep Learning, Machine Learning, Artificial Intelligence (AI), Apache Airflow, SQL, Docker, Matplotlib, NumPy

Computer Vision Engineer

2015 - 2018
Akiyama
  • Developed a computer vision library for face analysis targeting passport compliance (ICAO).
  • Developed image processing algorithms such as white balance adjustment and motion stability analysis for automatic capture and cropping of the image in the face area.
  • Explored a range of solutions for face segmentation, including deep learning, segmentation models, motion analysis, and Intel RealSense RGBD cameras.
  • Developed algorithms for fingerprint matching under robust image scale variance caused by the individual aging as part of the solution to help find missing children.
  • Built a fully convolutional neural network (FCN) classifier and regressor to compute fingerprint segmentation and scale estimation.
  • Developed an application for fingerprint data acquisition using custom fingerprint hardware.
Technologies: DevOps, Augmentor, Dlib, CMake, AI Design, Canon, Convolutional Neural Networks (CNN), Intel RealSense, Emotion Recognition, Facial Recognition, Biometrics, Image Processing, NoSQL, Embedded Hardware, Microsoft Kinect, Git, TensorFlow, Pandas, Python, C++, Linux, Deep Learning, Machine Learning, OpenCV, Artificial Intelligence (AI), Computer Vision, Julia, Matplotlib, Docker, Qt

Image Processing Filter

https://github.com/tibaes/coherentLine
An image filter to convert a photo into a set of coherent lines. This project was a university assignment and is a little out of date. However, it was challenging to implement following only the original paper, yet, the results are visually pleasant. The results were presented in the presentation PDF.

Scene Understating

This was academic research about motion analysis algorithms. I developed an object-tracking solution using object detection and optical flow to guide PTZ cameras to follow a given target. I created a classifier for actions in videos, using Gaussian Mixture Models of motion patterns based on optical flow and Kullback-Leibler divergence. I received three prizes for best undergraduate work at the Federal University of Parana (UFPR) events.
Technologies used: C/C++, Make, OpenCV, OpenCL, CUDA, and PTZ Cameras.

3D Reconstruction of Artifacts

This was an academic work, funded by UNESCO, to help with the preservation of statues and pieces of art in museums and churches. The project's core was the usage of high precision 3D scanners and high-resolution professional cameras to reconstruct the 3D and texture information of the artifacts. I acquired 3D/depth images of artifacts, studied 3D reconstruction using ICP and shape from motion techniques, and provided assistance to other researchers.
- Technologies used: Microsoft Kinect, AICON Breuckmann Scanner, Canon DSLR (CHDK), C++, OpenCV, and PCL.

3D Face Detection on GPU

My final project to complete my degree was about face detection on Microsoft Kinect depth images and GPU. I optimized a 3D face projection algorithm to compute on GPU.
Technologies used: Microsoft Kinect, C++, OpenCL, and CUDA.
2016 - 2019

Master's Degree in Computer Science

Federal University of Bahia (UFBA) - Salvador, BH, Brazil

2008 - 2012

Bachelor's Degree in Computer Science

Federal University of Parana (UFPR) - Curitiba, PR, Brazil

JANUARY 2018 - PRESENT

Machine Learning Engineer

Udacity

Libraries/APIs

OpenCV, TensorFlow, Scikit-learn, Pandas, NumPy, SciPy, PyTorch, REST APIs, Matplotlib, Dlib

Tools

Git, Jupyter, Seaborn, Jetson TX2, Plotly, Apache Airflow, CMake, Jira

Platforms

Jupyter Notebook, Linux, NVIDIA CUDA, Docker, MacOS, iOS, Raspberry Pi, Google Cloud Platform (GCP), Amazon Web Services (AWS), AWS Lambda

Industry Expertise

Insurance

Other

Computer Vision, Image Processing, Machine Learning, Artificial Intelligence (AI), Biometrics, AI Design, Data Analysis, Data Analytics, GPU Computing, Deep Learning, Credit Risk, Pricing, Convolutional Neural Networks (CNN), Video Processing, Facial Recognition, Emotion Recognition, PTZ Cameras, Intel RealSense, DataOps, 3D Image Processing, 3D Reconstruction, Embedded Hardware, Canon, Serverless, Augmentor

Languages

Python, C++, Swift, SQL, Julia, Rust

Frameworks

Microsoft Kinect, Qt, Flutter

Paradigms

Scrum, Data Science, DevOps, Microservices, ETL

Storage

Google Cloud, NoSQL

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring