Thinking Nine
Custom Rose Trail
x(t) = 50 + (6.8 cos t - 3.1s cos 8t) * 4.3 y(t) = 50 + (6.8 sin t - 3.1s sin 8t) * 4.3 s = detailScale(time)View All
Custom Rose Trail
x(t) = 50 + (6.8 cos t - 3.1s cos 8t) * 4.3 y(t) = 50 + (6.8 sin t - 3.1s sin 8t) * 4.3 s = detailScale(time)View All
A collection of projects that I have worked on.

Developed local RAG pipeline on 30GB mixed data (PDF, images, Word, Excel...) Markdown conversion via DeepSeek OCR and adaptive header-based segmentation, Mistral E5 embeddings Weaviate DB, Enriched metadata (page, subject, year, exercise type)

Team AI project on a from-scratch sequential Q*bert game: genetic algorithm baselines, Deep Q-Learning and Dual DQN agents with a convolutional network over 7×7 state tensors, plus reward shaping and a custom visualizer.

Built for an English class project: practice technical interviews with a local pipeline (Mistral Nemo, faster-whisper, Qwen3 TTS via FastAPI). Customize interviewer voice, kindness, and topic; optimized for natural back-and-forth speed.


Participated in the IA Kathon 2026 Nancy at ARTEM (we won), where we developed a chatbot leveraging LangGraph to standardize and process client commands efficiently for a B2B company specializing in cloud, telecom, and cybersecurity services.


The TIPE is a project of preparatory school graded at the final entrance exam where my subject was Nimber concept and applications

This portfolio website features a minimalist design, smooth scroll animations, contact form integration.

Training and preparation for SWERC 2025, results : 10/46 France, 41/143 Southwestern Europe