Skip to content
View Gael926's full-sized avatar

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Highlights

  • Pro

Block or report Gael926

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Gael926/README.md

What I'm working on

I work at ARQUUS (John Cockerill group) on a generative AI platform running on isolated, on-premise infrastructure: model serving, document retrieval, internal tooling, access control. Most of the interesting problems there are not about models. They are about making a system reliable and verifiable when calling an API is not an option.

Going deeper on three things through 2027: inference serving and where it actually breaks, retrieval evaluated with numbers instead of intuition, and writing MCP servers rather than only consuming them.

Before that: an OCR and LLM pipeline that processed 40,000 handwritten archive pages, and a full-stack PWA built on automation workflows and cascaded LLM calls.

Live demos

Credit scoring · Course completion · Portfolio optimisation · Tomato yield

Code and results in the pinned repositories below.

Stack

  • Models & data - Python, ML/DL (PyTorch, scikit-learn, LightGBM, XGBoost, SHAP, Optuna), SQL, Pandas, NumPy
  • LLM systems - MCP, RAG, Hugging Face, LM Studio, Ollama, OpenWebUI, OpenAI-compatible APIs, fine-tuning
  • Systems - Linux, Docker, Git, GitHub Actions, SSH, Bash, MLflow
  • Web & automation - JavaScript, FastAPI, n8n, Supabase

lereungael@gmail.com · Portfolio · LinkedIn · CV

Pinned Loading

  1. credit-scoring-analysis-pipeline credit-scoring-analysis-pipeline Public

    Data analysis & feature engineering on Home Credit dataset (30M+ rows, 10 files). ML pipeline with business cost optimization, SHAP explainability, and Streamlit dashboard.

    Python

  2. hackaton-sival-agreen-ctifl hackaton-sival-agreen-ctifl Public

    Greenhouse tomato yield prediction: ElasticNet vs TOMSIM physiological model. Agreen CTIFL hackathon.

    Jupyter Notebook

  3. dl-course-completion-prediction dl-course-completion-prediction Public

    Neural networks (PyTorch, TensorFlow) predicting student completion. Multi-target classification and regression with SHAP.

    Python

  4. portfolio-optimization-multi-criteria portfolio-optimization-multi-criteria Public

    Multi-objective portfolio optimisation with NSGA-II (return, risk, cost). Streamlit dashboard, benchmarked against Markowitz.

    Python

  5. ml-advanced-retail-segmentation ml-advanced-retail-segmentation Public

    End-to-end unsupervised learning pipeline (RFM, UMAP, K-Means) to segment 1M+ transactions into actionable customer personas.

    Jupyter Notebook

  6. ml-credit-card-fraud-classification ml-credit-card-fraud-classification Public

    Detecting credit card fraud using ML models (LogReg, RandomForest, XGBoost)

    Jupyter Notebook