Job Description

The world is more connected than ever before and the pace of change is bewildering. We deliver the best combination of technology and services to enable our clients to fulfil the diverse needs of their customers. By fostering a deeply and personal relationship with our clients, we want to improve the lives of people and organizations.

Axians is the VINCI Energies brand dedicated to ICT and Digital Transformation. We’re present across 38 countries, with more than 16 600 employees and an annual revenue of more than 3.7 billion euros, in 2024.

We are seeking a Lead Data Scientist to architect and build next-generation voice-enabled virtual assistants powered by large language models. This role will focus on creating seamless, natural conversational experiences that combine advanced LLMs with state-of-the-art voice AI technologies.

Key Responsibilities

• Lead the design and development of intelligent virtual assistants that integrate LLMs with real-time voice interaction capabilities
• Architect end-to-end conversational AI pipelines, including speech recognition, natural language understanding, LLM reasoning, response generation, and text-to-speech synthesis
• Optimize LLM performance for conversational contexts, including prompt engineering, fine-tuning, and context management for multi-turn dialogues
• Design and implement low-latency inference systems to enable responsive, human-like voice conversations
• Build evaluation frameworks to measure conversation quality, coherence, accuracy, and user satisfaction
• Integrate multi-modal capabilities allowing the assistant to process and respond to both voice and text inputs seamlessly

Technical Focus Areas

• LLM Integration Adapting large language models (GPT, Claude, Llama, etc.) for conversational contexts and virtual assistant use cases
• Conversational AI Dialogue management, context tracking, memory systems, and multi-turn conversation optimization
• Real-time Systems Low-latency streaming architectures for natural voice interactions with minimal delay
• Personalization Building adaptive systems that learn user preferences and communication styles

Required Qualifications

• PhD or Master's degree in Computer Science, AI, NLP, or related field with 4+ years of experience,
• Extensive experience working with large language models (GPT-4, Claude, Gemini, LLaMA, etc.) and prompt engineering
• Deep understanding of conversational AI architecture, including dialogue systems and natural language understanding
• Strong knowledge of transformer architectures and experience fine-tuning LLMs for specific domains
• Proficiency with modern ML frameworks (PyTorch, TensorFlow) and LLM APIs (OpenAI, Anthropic, etc.)
• Experience optimizing models for production, including latency reduction and cost optimization