➤ Bonjour ! I’m an ML engineer at an AI startup, working on NLP, model training, evaluation and data engineering.
➤ I study Mathematics & Computer Science at Université de Montréal.
➤ A few things I’ve built: CloudVisionServe — serving vision models • Interpreting Tabular Models — SHAP & PDP • Hilbert matrices — numerical stability
➤ Right now, I’m digging into GEPA, recursive language models (RLMs) and better LLM evaluation.
➤ Find me on Hugging Face • LinkedIn.
- Montreal
- in/mateo-coda-forno-19b475272
Pinned Loading
-
retinal-grading-under-imbalance
retinal-grading-under-imbalance PublicA comparative machine-learning study for a five-grade diabetic retinopathy Kaggle competition, investigating model selection and evaluation under class imbalance.
Jupyter Notebook
-
CloudVisionServe
CloudVisionServe PublicClassify uploaded images with four pretrained PyTorch/TorchVision models through Flask and Streamlit, packaged with Docker for Google Cloud Run.
Python
-
interpreting-tabular-models
interpreting-tabular-models PublicTwo Random Forest case studies connecting validation performance, global feature reliance, local SHAP attribution and input representation.
Jupyter Notebook
-
hilbert-matrix-numerical-stability
hilbert-matrix-numerical-stability PublicMATLAB experiment on ill-conditioned Hilbert systems: why residual, coefficient error and function error can tell different stories.
MATLAB 1
-
beer-quality-model-selection
beer-quality-model-selection PublicSeven-grade beer-quality classification on imbalanced Kaggle data, from KNN baselines to Bayesian-tuned HistGradientBoosting.
Jupyter Notebook
-
audioset-clip-reconstruction
audioset-clip-reconstruction PublicReconstructing label-selected audio clips from Google's AudioSet metadata with Python, yt-dlp, and FFmpeg.
Python
If the problem persists, check the GitHub status page or contact support.