Francisco Farinha

Francisco Farinha

Data Scientist @ BlackBerry

University of British Columbia

Hi,

I’m Francisco, a Data Scientist at BlackBerry who is passionate about tackling real world problems with the exciting potential of Machine Learning. Currently, my interests are in model security and differential privacy, as well as anomaly-based IDS.

While you’re here, check out some of my projects, view my resumé, or drop me a line below.

Interests

  • Machine Learning
  • Differential Privacy
  • Computer Vision
  • Data Analysis

Education

  • BASc in Engineering Physics, 2017 - 2023

    University of British Columbia

Experience

 
 
 
 
 

Data Scientist

BlackBerry

July 2023 – Present Waterloo, ON
  • Optimized IVY synthetic sensors, reducing CPU and memory usage to within 5% for ML-IDS solutions.
  • Partnered with the Threat Research and Intelligence team on automotive cybersecurity research, creating a new MITRE framework.
  • Conducted GPT-4 experiments for SOC analyst support, focusing on classification and generation of MITRE unit tests for IDS solutions.
  • Spearheaded research into differential privacy techniques to protect ML models against model inversion attacks, aiming to integrate these into an ‘ML on the Edge’ framework.
 
 
 
 
 

Data Science Student

BlackBerry

September 2022 – December 2022 Waterloo, ON
  • Assessed BlackBerry’s custom anomaly detection algorithm against standard Intrusion Detection Systems for feasibility.
  • Improved facial detection module speed during project escalations and educated BlackBerry Labs on facial recognition loss functions.
  • Developed and delivered a workshop on Convolutional Graph Neural Networks, exploring their use in malware classification.
 
 
 
 
 

Research Assistant

Canary Cognition, UBC

September 2021 – April 2022 Vancouver, BC
  • Training and fine-tuning BERT models on transcribed speech data for Alzheimer’s Disease classification.
  • Exploring visualization techniques for model interpretability to find patterns in input data.
  • Participating in weekly reading groups to present novel papers.
 
 
 
 
 

Machine Learning Specialist

Flash Forest Inc.

September 2020 – April 2021 Surrey, BC
  • Developed QGIS Machine Learning segmentation pipeline to aid in planting missions.
  • Compiled, cleaned, and maintained dataset of orthomosaic images.
 
 
 
 
 

Machine Learning Intern

Longervision Technology

July 2020 – November 2020 Surrey, BC
  • Prepared client image dataset trained and deployed a quantized YOLOv4 model on NVIDIA Jetson Nano for detecting passengers and workers on train platforms.
  • Established a Structure from Motion (SFM) pipeline for 3D reconstruction of drone footage using OpenMVG and OpenMVS.
 
 
 
 
 

Computational Plasma Engineer & Programmer

General Fusion Inc.

January 2019 – April 2019 Burnaby, BC
  • Integrated Magnetohydrodynamics stability framework OMFIT into physics workflow.
  • Developed additional functionality for OMFIT – visualization tools, PBS/Torque server compatibility, parallel job submission – which decreased timeline processing by over 100%.
  • Presented DCON Stability Analysis reports to the MHD team weekly.

Projects

*

GAN Who

Guess Who implementation with GAN generated and VDSR upscaled images.

NumPy Neural Networks

Implementation of Neural Networks using only NumPy

UBC Open Robotics

Developing Software to compete in the RoboCup@Home competition.

Machine Learning Competition

Implemeted YOLO to navigate a simulated course for ENPH 353.

Artifact Removal & Biomarker Segmentation

A Project for EECE 571T - Advanced Machine Learning Tools - Where I created a pipeline to detect FOXP3+ biomarkers in follicular lymphoma TMA cores.

Paper Review

A Neural Algorithm of Artistic Style

Creating artistic images using Deep Neural Networks

Accurate Image Super-Resolution Using Very Deep Convolutional Networks

Using a Deep CNN to achieve highly accurate single-image super-resolution

Contact