Applied ML / ML Software Engineer · Montréal
|
PyTorch · Python · C++ · ML pipelines · Production tooling.
Machine Learning Engineer with applied research in deep learning, computer vision, and reinforcement learning. Builds, trains, and evaluates neural networks — 3D segmentation, object detection, deep RL agents.
I back that modeling with 3+ years building and supporting production tooling and pipelines at DNEG and Ubisoft Montreal.
Open to Applied ML / ML Engineering roles in Montréal or remote across Canada.
Professional Experience
Research Engineer | ML & operations research
2025 – PresentÉTS — Numerix Lab
Conducted research on applications of Deep Reinforcement Learning in construction project scheduling.
Software Engineer | Software Testing & Tooling
2024 – 2025Ubisoft Montreal
Conducted testing of editor builds and internal tools. Collaborated with software engineers and artists to reproduce critical issues, prioritize fixes, and improve overall software stability.
Software Engineer | Tools & Infrastructure
2022 – 2024DNEG
Developed tools for industry-standard software including Maya, Clarisse, Houdini, and Nuke. Collaborated with artists and technical directors to enhance DCCs and proprietary tools.
Teaching Assistant
2021École de technologie supérieure (ÉTS)
Course: GTI320 Mathematical Programming (2 sessions). Presented assignments, answered questions on computer graphics and C++ debugging, and graded assignments.
Mitacs Globalink Research Intern
Summer 2019Université de Moncton
Developed a convolutional neural network (U-Net architecture) for 3D semantic segmentation of lung CT scans to detect tumors. Collaborated with the research team to secure first place in an industrial problem-solving competition.
Selected Projects
Artificial Intelligence & Machine Learning
Wellbore Geology Prediction for Geosteering
Kaggle competition solution that reconstructs the True Vertical Thickness (TVT) of horizontal wells for geosteering by cross-correlating gamma-ray logs against a vertical reference "typewell". Engineered well-relative features (GR cross-correlation, per-well calibration) and combined CatBoost, LightGBM, and XGBoost models via greedy hill-climbing ensemble selection, validated with GroupKFold across 773 wells.
Sentiment & Emotion Analysis
An NLP and computer-vision project exploring two ways to read emotion: a text-sentiment track benchmarking the classic NLTK VADER lexicon against a fine-tuned Twitter-RoBERTa transformer on the TweetEval dataset, and an image/video emotion track built on a CLIP + CoOp-style learnable-prompt architecture with a temporal transformer over per-frame features. Ships a Gradio web UI to test both models.
Medical Image Segmentation with Deep Learning
Developed a 3D-UNet framework for lung cancer tumor segmentation from CT scans during a Mitacs Globalink research internship at Université de Moncton (2019). Achieved 92.1% sensitivity in tumor detection.
Evolutionary Algorithms for Combinatorial Optimization
Implemented genetic algorithms for solving the Knapsack Problem, exploring evolutionary computation techniques including selection, crossover, and mutation operators.
Real-Time Video Surveillance Event Detection
Built a real-time intrusion detection system for video surveillance using Tiny-YOLO-VOC via Darkflow, OpenCV, and TensorFlow (2018). Users define protected zones on live webcam feeds, and the system triggers alerts when persons are detected within monitored areas using IOU-based overlap analysis.
Computer Graphics & Simulation
Ray Tracer from Scratch
Personal exploration of ray tracing fundamentals through a C++ implementation following Peter Shirley's Rendering series. Features CPU-based rendering with multiple material types.
Advanced Ray Tracing Renderer
CPU-based ray tracing renderer in Java with OpenGL integration. Features geometric primitive intersection, anti-aliasing, depth of field, and texture mapping.
Physics-Based Character Animation
Physics-based animation system using Proportional-Derivative (PD) controllers. Simulates a skeletal puppet suspended by virtual marionette strings in Java/OpenGL.
Finite Element Method Simulation
2D finite element method simulation for deformable bodies with implicit backward Euler time integration for numerical stability.
Embedded Systems
Smart Irrigation System
IoT-based smart irrigation system using temperature, humidity, and luminosity sensors. Implemented in C on an STM32F302R8 microcontroller.
Technical Expertise
Education
M.A.Sc. in Information Technology
École de technologie supérieure (ÉTS)Thesis: Hyperparameter tuning for the projected Gauss-Seidel method in rigid body simulations
B.Eng. in General Engineering
Tunisia Polytechnic School
Preparatory Classes (Mathematics–Physics)
Tunis Preparatory Engineering Institute (IPEIT)Certifications
Professional Machine Learning Engineer In Progress
Google CloudIndustry certification covering the design, training, and productionization of ML models and pipelines on Google Cloud.
Drone Pilot Certificate — Complex Operations Level 1 (OCN1)
Transport CanadaAuthorizes Beyond Visual Line of Sight (BVLOS) remotely piloted aircraft system (RPAS) operations in low-risk environments.
Contact
Open to Applied ML / ML Engineering roles in Canada. Let's talk.