×
Loic Mandine

Loic Mandine

Machine Learning Engineer

Montreal, Canada, Quebec
+33 6 85 36 16 69
French, English

Background


About

About

Seeking a full-time position in the Machine Learning field, with a broad interest in AI applications and real-world problem solving.

Work Experience

Work Experience

  • Machine Learning InternHydro-Québec Research Institute (IREQ)

    May, 2024 - Jan, 20258 months

    Automated reading of gauges/indicators from various models of electrical equipment.

    • Built and deployed an automated gauge-reading pipeline (integrated into ROS for the inspection robot) to interpret real-world electrical equipment indicators.

    • Contributed to Vision-Language Model (VLM) fine-tuning, handling data preprocessing and annotation with very limited samples to improve gauge understanding.

    • Designed a modular, extensible framework with integrated visualization tools, enabling rapid experimentation and future improvements.

    • Established a structured, annotated data pipeline from robot inspection missions, supporting continuous model training and validation.

  • Machine Learning EngineerMARSS Group

    Dec, 2022 - Jul, 20237 months

    Develop/Implement production-ready Object Detector model (mainly one-stage CNNs) & Compress those lasts for edge low-power devices deployment

    • Deep Learning to the edge

    • Computer Vision - Object Detection

    • Neural Network Compression

  • Data Scientist ApprenticeshipAzurIA

    Sep, 2021 - Nov, 20221 year 2 months

    Still at its very beginning, I joined AzurIA to support the R&D-team's AI projects. I had the opportunity to work on many Computer Vision-related projects.

    • Life of the creation of a startup

    • Computer Vision - Satellite Imagery/Remote Sensing, Cloud Segmentation (Custom U-Net), Boat Detection (YOLOvx model series)

    • Literature Survey, State-of-the-art of DL datasets/algorithms applied to multi-spectral images

    • Anomaly Detection - Defect detection manufactured pieces, Disease Segmentation on vign leafs

    • DevOps/MLOps - Self-hosted ML workspace & GPU cluster maintenance, Gitlab CI/CD, PoC of ML projects

    • TinyML - Deployment ML models on edge devices (several nvidia jetsons, raspberry pi/pico, intel Movidius/OpenVino toolkit)

  • Data Scientist InternCenter Antoine Lacassagne, CHU Nice - Department of Epidemiology, Biostatistics and Health Data

    Jun, 2021 - Aug, 20212 months

    This internship was to develop and validate a machine learning model to help in the decision of malignancy of a thyroid nodule. Then, the performance of several machine learning methods were compared.

    • Machine learning for medical diagnosis in oncology

    • Publication of a scientific article in a peer-reviewed journal

Skills

Skills

  • Machine Learning

    Pytorch

    Pytorch Lightning

    OpenMMLab libs

    Wandb/Comet/ClearML

  • Software Engineering

    Git

    Docker

    Remote Development

    Wandb/Comet/ClearML

    Pandas/ScikitLearn

  • Linux & System Administration

    Shell scripting (Bash)

    Ansible / Infrastructure as Code

    Docker & container orchestration

    CI/CD (GitLab CI, GitHub Actions)

    Networking (SSH, VPN, proxies, firewalls)

    Monitoring & logging (Prometheus, Grafana)

    Self-hosted services & server maintenance

  • Soft Skills

    Strong communication skills

    Adaptable & proactive learner

    Open-minded and receptive to feedback

    No ego / approachable

    Positive and supportive mindset

Education

Education

  • Master in Informatics - Specialization in Machine Learning, Master, MILA Institute / University of Montreal

    Sep, 2023 - Sep, 2025

    Representation Learning

    Machine Learning Fundings

    Data Science

    Advanced Deep Learning projects

    AI engineering

  • Applied Mathematics and Modeling - Data Science, Baccalaureate + Master's degree, Polytech Nice Sophia-Antipolis

    Sep, 2017 - Sep, 2022

Certificates

Certificates

Awards

Awards

  • Montreal Robotic Summer School 2024 - 1st place , MILA Institute

    Awarded on: Jun 14, 2024

    We tested PPO and Student-Teacher Distillation (+ Random Network Distillation (RND) & Symmetry-based Augmentation) to train the Unitree Go1 robot to walk on rough terrain using NVIDIA's Isaac Gym.

  • 1st place - AI Challenge , Hack Quebec 2024

    Awarded on: Mar 17, 2024

    We developed a project that leverages AI to detect snow and ice accumulation on power lines using an unlabeled dataset of 235,000 images. A robust ensemble approach—combining SSIM, k-means, the open-source vision-language model LLaVA, and a convolutional neural network—assigns graded accumulation labels while reducing false negatives through weighted voting.

  • Hackaton - 2nd place , BonsAPPs/BonsEyes european organization

    Awarded on: Apr 01, 2022

    AzurIA R&D team has successfully participated in Stage 1 of the 1st support programme of BonsAPPs. Throughout a two-month program, we developed a structured datatool and an AI Asset workflow for the training and deployment of deep neural network on the edge for a Defect Detection usecase.

Volunteer Work

Volunteer Work

  • System Administrator, University of Montreal - DIRO Department

    Feb, 2025 - Feb, 2025

    Preparing a multiplayer gaming experience around the Flightgear simulator and challenges around this game.

    • Configured and deployed multiplayer servers for the FlightGear flight simulator, enabling participants to compete in custom aviation challenges.

    • Built streaming servers with Moonlight/Sunshine to deliver low-latency remote gameplay experiences.

    • Used Ansible and Docker to automate deployment of client and server environments for the FlightGear multiplayer setup.

    • Integrated Chisel to proxy network connections, ensuring secure and reliable access across environments.

    • Prepared custom challenges using .kml files and the FlightGear Python SDK, creating mission-based scenarios for hackathon participants.

    • Managed networking, scalability, and technical support to guarantee smooth operation throughout the event.

  • Tutor, University of Montreal - DIRO Department

    Sep, 2023 - Jan, 2024

    Animator of weekly seminar on 'Learning Linux' for students. I also help (undergrad) students with their homeworks and projects.

    • Linux Seminar for students

    • Projects/Homework assistance, tutoring

Publications

Publications

Interests

Interests

  • Biomimicry

    BiomimetismBio-inspired innovation
  • Open Source & DIY Tech

    Open source contributionLinux tinkering & self-hostingDIY electronics & Raspberry Pi projects3D printing & hardware hacking