Data Science & Machine Learning · EPITA SCIA 2026 · Paris, France
Engineer first, I care about maintainable code.
I build ML systems end to end, from RAG pipelines and recommender models to
distributed orchestration and real-time trading automation.
-
KairosEdge — Crypto Arbitrage Toolbox · personal · 2023 → today
Modular Python toolbox automating Cash & Carry and funding-rate arbitrage across spot, futures and perpetual markets, built around a templater design pattern with live Bybit/Deribit API integration and spread monitoring. -
KuaiRec Video Recommender
Preference prediction on the KuaiRec dataset — distributed ALS and a Two-Tower neural network in Keras. -
Poetry Classification & Generation
Benchmark over a 60k+ multilingual poem dataset comparing 10+ methods, from classical ML to RNN-with-attention and GPT-2. -
Poisoned Teacher & CoT Distillation
Benchmark of backdoor transfer through knowledge distillation, quantifying the Clean-Accuracy vs Attack-Success-Rate trade-off across model scales and poison ratios. -
POSIX Shell in C · best team of the generation
POSIX-compliant shell with a lexer-parser architecture supporting expansion, piping, subshells, redirections and built-in commands.
ML & Deep Learning
LLM & Retrieval
MLOps & Cloud
Automation & Scraping
