about me
I do
Data Science,
ML, and
Software Engineering
and I look for ways to use technology for social good
education
MSc at the University of Porto (FEUP), in 2020.
Spent a semester at the University of Edinburgh as a graduate student,
and another at the University of Toronto developing my MSc thesis:
"High-level Approaches to Detect Malicious Political Activity on Twitter".
work & more
Creator of Desarquivo and Twitter Watch.
Worked at
Wise,
LIAAD,
I3S,
SPECS.
Previous speaker at
QCon London,
Future of Computing,
Pixelscamp,
WiE ILS,
IBM Qiskit Camp.
projects & tools
Here are some public projects that I have developed overtime, new ones every now and then
Hackacity Porto 2019
Improve decision making over the local housing problem by using: scraping of booking.com, scripts to gather Porto's open data, cleanup and dataset building, regression techniques to infer which features are more important, clustering, ...
π¬bioseq
A python module for biological sequencing, includes most operations on DNA/RNA/Protein sequences, local and global alignment, BLAST, multiple sequence alignment, UPGMA, phylogenetic trees, similarity graphs, ... Developed for the Algorithms for Bioinformatics course. 100% code coverage.
STUNS (Structring The UNStructured)
Command line tool to read raw datafiles collected from mobile phone sensors, report data quality metrics via HTLM reports, insert organised information into NoSQL Database. Expose new queries through flask REST API. Developed for Fraunhofer
π Pyhparser
Pyhparser is a Python module that parses input files input by describing the format the data is in, and allowing you to use them as your own variables. Due to its simplicity and speed it is great for Hackathons, Machine Learning and day-to-day data tasks.
I built it to save time in Google Hash Code.
πΈ feup-iart
Artificial Intelligence projectMuticlass classification of Frog Species using Deep Neural Networks in TensorFlow, based on their calls π
Check out the Jupyter Notebook with the source code.
The data is available on a UCI dataset.
π½ LAT - Lara Autotuning Tool
- Developed at SPeCS at FEUP.
- LAT is a tool for code autotuning, built using LARA (a Javascript based language that supports source code transformations) and applied using the Clava tool.
- LAT mimics Intel Software Autotuning Tool (ISAT) behaviour for testing multiple instances of the same code. However, it is build entirely on LARA. Because of this it is platform agnostic and can be more flexible and easier to expand, both in functionalities and languages (not only C and C++).
π SigTools
Sigarra Tools - Sigarra on Steroids: Productivity Tool to Export Calendar Events, Infinite Scroll, DataTables, and more (SigToCa's heir). Available on the Google Web Store, Firefox Add-ons or even direct import from the .crx.
πΊ feup-plog
Logic Programming Course, consisting of two projects:- Project 1: Prolog implementation of the LYNGK board game (logic and command line interface) - report (pt)
- Project 2: Constraint Satisfaction Problem (clpfd) consisting of a "Constraint Logic Programming Approach to Teacher Hour Allocation for University Subjects" - paper
- Extra 1: Includes code for Logic Programming exercises as well as code for 20+ exam solutions
- Extra 2: Python was used to develop a tool to automatically generate, parse and test mathematically valid problem instances.
π feup-cal
Algorithm Design and Analysis
Garbage Truck Route Generation using real OpenStreetMaps data and Google Maps javascript API. Implementation of many graph algorithms by taking advantage of many C/C++ features like: Classes, Templates, Operator Overrides, ...
feup-ltw
Web Technologies Laboratory
A TODO list management platform, the JavaScript interface, including AJAX and events was done through pure JavaScript (intentionally). I actually built a primitive type of Query Builder inspired by Laravel's Queries.
Matrix - Bank Security Matrix Card Management
Real World Android Application to manage Bank Security Matrices: Minimalist, efficient, data is encrypted and the app is password protected. This allows people to stop carrying paper copies of their security matrices.feup-sdis
Distributed SystemsDistributed P2P File Backup System. Desgin and implementation of a new Peer Clustering Algorithm (see full description)
feup-cgra
Computer Graphics
A WebGL (through
WebCGF) project consisting of a user-controlable submarine.
feup-rcom
Computer Networks- Implementation of a Communication Protocol (Data-Link Layer - 2 in OSI Model)
- Implementation of an FTP client
- Full Network Configuration with Router and Switch
feup-sope
Operating SystemsA tool that mimicks Unix's find along with a Sauna Client Generation and Management system. Interesting Features:
- Signal and Signal Handling
- Multiprocessing and Multithreading
- Command line arguments
- Pipes, FIFOs, Mutexes, semaphores
feup-aeda
Algorithms and Data Structures ProjectBookings Management System - a proof of concept implemented in C++. Using BSTs and HashTables.