Freelancer profile translated to English.

Description

My talent: Designing systems where language models (LLMs), behavioral data, and user experience unite to create intelligent, explainable, and truly useful products.

How?

I develop agentic LLM architectures (via MCP, WebSocket, REST/SSE) capable of analyzing, recommending, and acting, with fine-tuning, RAG, dynamic prompts, and UX data visualization logic.

Specializations:

• Fine-tuning open-source LLMs (DeepSeek, Mistral…) with Unsloth / QLoRA / FSDP

• Orchestrating AI agents via MCP Server, dynamic prompts, WebSocket, SSE

• Building market analysis engines through sector-specific RAG (LLaMaIndex, Deepsearch)

• Data UX integration: user behaviors, clustering, forest models, strategic insights

Why work together?

• Because I know how to articulate the rigor of AI systems (MCP, FlashAttention, QLoRA, GGUF…) with the richness of human experience (Design Ops, UX Research, AI Thinking).

• Because I design AIs that understand, contextualize, and adapt, not just models that respond.

• Because I develop AI products designed to be used, understood, and loved.

You need...

• …LLMs for intelligent journeys?

• …Data + UX to model agentic interactions?

• …Operational, scalable, and humanly relevant AI systems?

Let's discuss. I am designing the AIs we will love to use tomorrow, today.

Industry field of expertise

Languages

English
Fluent
French
Native or bilingual
Hebrew
Conversational

Workplace preferences

Can work on-site

Paris (up to 50km), Paris (up to 50km)

Dasein
President (MCP Server, Gen AI, Agentic AI, AI Design)
March 2024 - Today (2 years and 3 months)
Paris, France
• Development of a specialized business LLM (analysis, expert sourcing, matchmaking)
• Generation of contextualized strategic advice through prompt engineering
• AI Market Research Engine: sector analysis + real-time data cross-referencing
• Agentic WebSocket Server: parallel execution of agent chains + dynamic prompt orchestration
• REST API and real-time inference via FlashAttention 2
• Development of a modular MCP Server, dynamically exposing business tools via:
• Secure REST endpoints (HTTPS)
• SSE (Server-Sent Events) for bidirectional streaming communication
• No-code Gradio interface for prototyping, orchestration, and agent testing
Infopro Digital
Lead R&D (Gen AI, Agentic AI, AI Design, AI Thinking)
June 2022 - March 2025 (2 years and 9 months)
Objective: Integrate Gen AI concretely and efficiently into digital products, by aligning user experience and industrialization of AI components.
Achievements:
• Evangelizing Gen AI to product, tech, and design teams
• Designing an AI Wireframe Kit: an AI-first standard integrating UX prompts, conversational flows, and agentic logic
• Integrating accessible Gen AI (A11Y + UX) into internal tools, facilitating cross-team adoption
• Formalizing AI ↔ user ↔ system interactions, foreshadowing the MCP agent chains now in production
• Launching an AI-compatible Design System (APY) structured according to LLM agent patterns
• Redesigning key platforms (Webikeo, Infopro, Automotive, Public Procurement), prepared for real-time AI component integration (APIs, embeddings, dynamic prompts)
Niji
UX Lead (ML, Data Analysis, UX Research)
June 2021 - June 2022 (1 year)
Paris, France
• Design of mobile apps (Barrière KYC, Betclic, Total EV) and redesign of key websites (Dalkia, RATP, Audencia…)
• Integration of data visualization (dashboards, UX KPIs) and user analysis via Python + Jupyter Notebooks
• Use of algorithms like Random Forest / clustering to identify UX patterns and optimize flows
• Contribution to the exploration of AI solutions applied to user experience