Lourenço Matos

Lourenço Matos

Researcher & Engineer

University of Cambridge

Biography

Aspiring researcher. Currently interested in Geometric Deep Learning and Neural Reasoning, working towards more reliable and generalizable deep learning. Current research spans Neural Algorithmic Reasoning applications to RNA velocity and Mechanistic Interpretability of LLMs through GDL principles. Also interested in the deployment and optimization of machine learning systems.

Interests
  • Geometric Deep Learning
  • Machine Learning Systems
  • Neural Algorithmic Reasoning
Education

  • MPhil in Advanced Computer Science, 2024 - 2025

    University of Cambridge

  • BSc in Computer Science and Engineering, 2021 - 2024

    Instituto Superior Técnico - University of Lisbon

Experience

 
 
 
 
 
PyTorch Contributor
PyTorch
February 2024 – July 2024
Implementing lowerings for torch.compile, optimizing average pooling backwards 3D by 100%.
 
 
 
 
 
Student Researcher
INESC-ID/IST
February 2023 – July 2024
Developing software tools for computational biology – boolean logical modeling of biological regulatory networks.
 
 
 
 
 
Volunteer Instructor
TreeTree2
July 2023 – July 2024
Instructor in online After School programs for young students in multiple scientific areas.
 
 
 
 
 
STT - Cybersecurity CTF Team Member
Instituto Superior Técnico
September 2022 – September 2023
Competing in cybersecurity Capture The Flag competitions.
 
 
 
 
 
Volunteer Student Mentor
Instituto Superior Técnico
October 2022 – July 2024
Student mentor of BSc and Erasmus MSc students.

Awards

  • “Lacaixa” Foundation Postgraduate Fellow (2024-2025)
  • Linux Foundation Scholarship - PyTorch (2024)
  • University of Lisbon Best Student Award (2021-2022)
  • Santander Excellence Scholarship (2023-2024)
  • Academic Excellence Diploma (2022-2023)
  • Oeiras Scholarship (2022-2023)
  • Academic Merit Diploma (2021-2022)
  • Oeiras Scholarship (2021-2022)
  • Secondary Education in Music with Merit (2008-2021)
  • Secondary Education with Merit (2018-2021)
  • Honourable Mention in Guitar Young Talents Contest (2012)

Certificates

  • Machine Learning Engineering for Production (MLOps) - DeepLearning.AI
  • Deep Learning Specialization - DeepLearning.AI
  • CEFR C2 English - Cambridge Assessment English
  • Open Water Scuba Diver
  • CS50x
  • Blockchain Principles

Side Projects

Avalanche Prediction

Bayesian approach to predict the severity of an avalanche in a given region, and identification of the optimal location for dam placement and avalanche mitigation. Bayesian Optimization, Multi-fidelity, Sensitivity Analysis, Gaussian Processes, etc.

Trading Bot

A machine learning pipeline for prediction of asset value using a customly designed LSTM model. Developed efficient pipelines for API data extraction, web-scraping, cleaning, and preprocessing data from hundreds of sources. All-time-high accuracy of about 64%. PyTorch, Tensorflow, Keras, Numpy, Pandas.

MyFriend

Web application allowing for the creation and conversation with an AI-generated psychologist. LLMs, Speech Synthesis and Voice recognition to create a realistic therapist character and enable a fluid conversation with the user. Node.js, React, AWS.

Compiler

Coded a compiler for this made-up language called TIL supporting recursion, conditionals, loops, etc. From scratch (no libraries), in C++.

Filesystem

Implemented a filesystem supporting concurrent access, multithreaded operations, and implemented a message broker on top of it. From scratch (no libraries), in C.

Bimaru AI Engine

Solved the Bimaru puzzle, finding the optimal solution extremely fast using search algorithms.