I'm a Software Engineering & AI student at EPITA, interested in building real-world systems at the intersection of AI, full-stack development, robotics, and data engineering.
I’ve worked on production SaaS products, AI-driven applications, financial data platforms, and humanoid robotics projects.
- 🎓 Engineering student at EPITA, specializing in Artificial Intelligence
- 🤖 Working on humanoid robotics & AI with Bouygues Group's Central Innovation Department
- 💻 Former Full-Stack Developer Intern at Padoa
- 🏆 Winner of the HEC × Stanford × Sciences Po Fintech / GenAI Hackathon
- 🥉 Finalist at the HRFlow AI Hackathon
- 🌍 French & Czech native, English C1
- 📍 Based in Paris, France
Languages
Python TypeScript Java C++ C
Web & Backend
Angular Node.js PostgreSQL
AI & Data
PyTorch Computer Vision Machine Learning
Tools & Platforms
Git Docker Power BI Excel SAP Business Objects ROS
Sep 2025 – Jan 2026 · Paris
Worked on a production HealthTech SaaS platform using Angular, TypeScript, Node.js, and PostgreSQL.
- Delivered scalable customer-facing features
- Built a configurable notification workflow with automated reminders and delivery tracking
- Designed scheduling logic supporting regulatory appointment constraints
- Collaborated closely with Product, Design, and Engineering teams
Jun 2023 – Dec 2023 · Paris
- Built internal web tools with JavaScript and Node.js
- Improved accessibility across web applications following WCAG principles
- Analyzed BI and data visualization platforms to support product strategy
Python PyTorch Transformers BPE CUDA
Built a decoder-only Transformer language model from scratch and trained it on TinyStories, covering the full path from tokenization to text generation.
- Implemented BPE tokenization, RoPE positional embeddings, multi-head causal attention, RMSNorm, residual blocks, and an MLP
- Built an AdamW training pipeline with gradient accumulation, BF16 mixed precision, validation, and checkpointing
- Trained an 8-layer, 8-head model for 20,000 steps on an RTX 4090, reaching 1.1627 validation loss
- Added greedy, temperature, and top-k decoding, plus a local web interface for testing prompts
- Implemented KV-cache inference and benchmarked 1.35× faster generation than naive decoding
Python RAG FAISS BM25 Mistral AI Streamlit
Evaluated Retrieval-Augmented Generation system for financial question answering on FinQA, focused on cited evidence retrieval and numerical reasoning.
- Built a full RAG pipeline with structure-aware chunking, FAISS dense retrieval, BM25, hybrid retrieval, reranking, and adaptive routing
- Improved numerical match from 47.37% to 61.11% compared with the dense baseline
- Reduced insufficient-evidence rate from 36.84% to 16.67%
- Increased calculation execution success from 52.63% to 72.22%
- Achieved 100% citation validity and 0% execution error rate on the latest documented 20-example adaptive reranked run
Bouygues Group · 2026 – Present
Python ROS Computer Vision AI
Exploring enterprise applications of humanoid robotics with Bouygues Group's Central Innovation Department.
- Developing AI applications for the AgiBot X2 Ultra
- Exploring computer vision and embedded intelligence
- Evaluating automation and human-robot interaction use cases
Scala Kafka Apache Spark PostgreSQL MinIO Docker
Distributed smart-grid monitoring platform combining real-time alerting with a batch analytics data lake.
- Built a Kafka-based telemetry pipeline for simulated transformer sensors
- Implemented alert detection, API handling, persistence, and notification workflows
- Designed Bronze, Silver, and Gold data lake jobs with Spark for historical analytics
HEC × Stanford × Sciences Po Fintech / GenAI Hackathon — Winner
Built an AI solution for predicting assembly-line breakdowns, with the goal of optimizing MRO procurement and reducing industrial downtime.
HRFlow Hackathon — Finalist
Developed an AI-first recruitment solution and helped secure a strategic partnership focused on data sovereignty.
Master in Engineering · Artificial Intelligence · 2021–2027
Top 10% admitted.
Relevant coursework:
Machine Learning · Data Engineering · Optimization · Statistics · Cloud Computing
Exchange Program · 2024
🇫🇷 French — Native 🇨🇿 Czech — Native 🇬🇧 English — C1 🇩🇪 German — B1 🇪🇸 Spanish — B1
Finance · Aviation · Rowing · Golf · Chess