Gabriel Fernandes Luz's Projects
This repository shows how we can use Tensorflow 2.X advanced functionalities in order to build our custom Loss Functions, Layers, Models and even Callbacks. This is very useful to adapt any given project to its own nature given business context and organization necessities.
This repository gathers the scripts you need in order to build and deploy a DAG to Apache Airflow, which automates the extraction of soccer matches results and then the loading of the acquired data into a database.
This repository will attempt to put in place an end-to-end ML development, with CI/CD capabilities. Basically, a model for fraud detection will be trained and then dumped to an Amazon S3 bucket. Hence, we will build a Docker container that will be automatically pushed do AWS ECR for further deployment with AWS Sagemaker.
This repository concerns with the steps needed to deploy a model in cases where the memory size restriction would be a problem. Hence, we can execute the lambda function in a containerized runtime, where all the dependencies are correctly set according to a Docker Image saved on AWS ECR.
Transfer learning with a Portuguese BERT model and further deployment in AWS SageMaker. I used the approach "bring your own code". Therefore, everything was developed with Docker and my own server for deployment.
Convolutional Neural Network: identifying cat and dog images.
This repository gathers some code required for estimating Average Treatment Effect (ATE) and Conditional Average Treatment Effect (CATE). On top of it, CI/CD capabilities will be added to the repository, so that it is possible to build a Python package for common causal inference frameworks.
This project aims to build an object identification algorithm, based on YOLOv3 in DarkNet framework, which is able to identify cigarettes and cigars.
Flask application blueprint for a social blog. Project deals with key concepts, such as authentication, forms and databases.
This repository aims at building a packaged solution for credit risk assessment. The main idea is, in accordance to Basel II, develop models for the A-IRB approach - in other words, internally estimating probability of default (PD), loss given default (LGD) and exposure at default (EAD), in order to compute expected loss (EL).
This repository gathers all the code required for an agent that uses Deep Deterministic Policy Gradient, in order to interact with a MuJoCo environment.
A blueprint for flask web applications.
Google IT Automation with Python Professional Certificate - Practice files
This is a library developed to incorporate useful steps in every data science and machine learning project, in order to facilitate and accelerate model development. Therefore, data scientists can spend less time working on coding preprocessing methods/scripts and use this time more wisely to create new features and tune the best model.
A collection of machine learning examples and tutorials.
Just a quick repo for getting snippets when necessary.
Testing functionalities for chatbot project
This repository aims at building recommender systems - from classical to most advanced approaches - with the full movielens dataset.
Repository for some techniques applied to reinforcement learning.
This repository is intended to perform Optical Character Recognition in a Brazilian ID document.
This repository focuses on providing an optimization solution to Job Shop problem, along with an interactive interface (dashboard) where the user can add jobs and restrictions, according to their business use case, and then get the results in a user-friendly fashion.
Classification problem: Identifying spam text messages.
Training and deploying a custom scikit-learn model using AWS Sagemaker.
Useful concepts and procedures on time series problems.