I make machine learning hold up on real equipment.
My work spans slow-leak detection for high-voltage circuit breakers, ML pipelines over 600K+ materials, and LLM agents a food plant uses every day.
- M.Sc. Computer Science (Machine Learning), McGill, December 2026
- Open to ML, applied AI and software engineering roles
- Based in Montréal · Canada or US (TN-eligible) · In person or remote, open to relocation
athmanebenarous.com · CV (EN) · CV (FR) · LinkedIn
Each title links to a short case study.
- Slow-leak detection for high-voltage breakers · Hydro-Québec Research Institute (IREQ), 2026
Sensitivity 0.867 vs 0.733 for the physical baseline on synthetic leaks, at an equal false-event budget. The gain is conditional, and the model is not certified for operational use. - LLM agents in production at a food plant · Russet House, 2026–present
Agents over ERP data, SharePoint and Teams, in daily use by plant staff since January 2026. - Self-learning line-speed advisory · Russet House, 2026–present
One recommendation for the input speed operators control, learned online from live SCADA data; operators accept, modify or refuse it. - ML pipelines over 600K+ materials · Chemia Discovery, 2026
Pipeline runtime from 30+ hours to under 1 hour. - 3D neuron soma segmentation platform · McGill capstone, 2024–present · code
One marker per soma (center and radius) gives its full 3D binary mask. - AlphaGateau for Breakthrough · McGill, 2026 · code
5×5 pretraining lifted the GNN's 8×8 win rate at iteration 5 from 17% to 67% (12-game greedy evaluation). - Energy profile of V-JEPA2 training · McGill, 2026 · code
Batch size 1 → 2: +42.4% throughput and −26.8% GPU energy per 1,000 samples.
Abstracts, PDFs and my part in each: athmanebenarous.com/research
- A Platform for the Semi-automatic Segmentation, Characteristic Shape Analysis and Simulation of Neuronal Somas (Sep 2024 – present)
- AlphaGateau for Breakthrough: A piece of cake? (May 2026)
- Energy Characterization of V-JEPA2 on the SOCS Teaching Cluster (Apr 2026)
- PolarBench: Multi-View Graph Construction and Benchmarking for Stance Detection in Decentralized Social Networks (Apr 2026)
- Optimizing Global River Flood Forecasting via Meta-Modeling (Dec 2025)
- Quantifying Temporal Change in Sentiment Analysis: A Study on the Amazon Reviews'23 Dataset (Mar 2025)
- ML: PyTorch, JAX, scikit-learn, TensorFlow, NumPy, pandas, GNNs, time series and anomaly detection, leakage-safe evaluation, RL (MCTS), LLM agents
- Data and infrastructure: Azure ML, AWS S3, Slurm GPU clusters, SQL Server, PostgreSQL, MongoDB, Git
- Software: Python, SQL, C, C++, Java, TypeScript, JavaScript, C#, Bash, React, Next.js, NestJS, Spring
- Industrial: Ignition SCADA, Modbus/OPC, Power BI, Power Automate, Power Apps
Every project, with dates and links, is on athmanebenarous.com.



