Projects

Tokenizer combining BPE with weighted multi-layer Transformer attention (v4) for semantically coherent merge decisions.

Independent search engine focused on privacy and transparency.

Family of three 15M-parameter French LLMs trained from scratch (no pretrained weights, no dependency on 🤗 Transformers) on a single GTX 1080 Ti, exploring looped…

HTML→Markdown extractor designed for LLM training corpora.

New security paradigm — Destructive Semantic Encoding (DSE) — distinct from classical cryptography, hashing, and compression.

Official terminal client for RDTvlokip Search, my independent, privacy-first French search engine (~5,021,846 indexed pages, in-house crawler).

Experimental testbed for whether a neural network can learn to produce language from a scalar reward alone, with random initial weights, no pretraining, and no input/output pairs.

An 11.05M-parameter residual network (128 channels × 8 blocks, recurrent depth ×2, gated memory, Squeeze-and-Excitation) that plays chess in one forward pass — no search, no…

A native technical markup language and VS Code extension, built entirely from scratch with zero runtime dependencies.

Collaboration with LMC.

Uptime monitor for Discord bots and your websites/APIs.