Senior AI and Machine Learning Scientist

16 ore fa

Firenze, Provincia di Firenze; Toscana, Italia team.blue Tempo pieno
Overview As Principal AI Solutions Engineer, you own end-to-end AI systems for a large European digital-services ecosystem. You set technical direction, ship production AI, and raise the bar for a fast-moving team. You work across model research, fine-tuning, agent-based orchestration, and real-time inference to power hosting, domains, email, and SaaS for millions of SMBs. You will shape architecture, guardrails, and reliable delivery with cross-functional partners. Architect and evolve multi-agent orchestration platform (Hermes / Multica) with plugin systems, tool pipelines, observability, and channel adapters Design and implement voice AI pipelines with low latency (STT, real-time TTS, VAD, SIP/RTP) Build and maintain RAG pipelines with retrieval quality measurement and hybrid search Define MCP server architecture and tool-use contracts across internal and third-party integrations Fine-tune and evaluate LLMs for domain tasks (customer support, classification, extraction) Evaluate model quality using automated and human metrics; manage model lifecycle from experiments to production Own AI observability stack (tracing, instrumentation, cost tracking, quality alerts) and enforce guardrails Develop data ingestion/preprocessing and feature pipelines; drive ML CI/CD, shadow deployments, canary releases Set architectural standards for AI systems; conduct design reviews; collaborate with Product/Commercial Engage with external research partners to identify signals for productionization 8+ years in ML Engineering, Applied AI, or Research Engineering with leadership experience ~ Deep hands-on experience with production LLMs: fine-tuning, RLHF/DPO, prompt engineering, RAG ~ Proficiency in Python and core ML stack: PyTorch, Transformers (HuggingFace), PEFT/LoRA ~ Experience with LLM inference serving in latency-sensitive environments (vLLM, TensorRT-LLM, or TGI) ~ Knowledge of agentic frameworks: multi-agent coordination, tool-use orchestration, observability ~ Experience with speech AI or real-time audio systems is a plus ~ Strong MLOps understanding: MLflow/W&B, model registries, Docker/Kubernetes, ML CI/CD ~ Awareness of LLM risk management (hallucination, prompt injection, data leakage, privacy) and mitigation ~ collaboration with cross-functional teams ~ voice AI pipelines (ASR, TTS, VAD, SIP/RTP) ~ RAG and hybrid search