Fullstack ai engineer
Pubblicato il 29-09-2026 - Azienda Anonima in Roma
What we are looking for?
A fast-growing Italian startup is looking for an experienced AI Engineer – Full Stack to join its team and contribute to the design and development of next-generation AI-powered products.
This role is ideal for someone who is comfortable building end-to-end Generative AI systems, from data ingestion and knowledge retrieval to production deployment, while collaborating effectively in a dynamic and fast-paced startup environment.
Key Responsibilities
Design, develop, and maintain LLM-based Generative AI pipelines, ensuring high response quality, reliability, and low latency.
Build and optimize Retrieval-Augmented Generation (RAG) solutions, including chunking strategies, embeddings, vector search, semantic retrieval, and re-ranking.
Design and implement Graph RAG architectures leveraging knowledge graphs and semantic relationships to improve reasoning and traceability.
Develop AI Agents and orchestrated workflows using function calling, tool usage, and multi-step reasoning.
Integrate leading commercial and open-source LLMs (e.g. OpenAI, Anthropic,
Google Gemini, Llama, Mistral), evaluating performance, quality, and cost trade-offs.
Design and develop scalable REST APIs and backend services exposing AI capabilities to frontend applications and third-party integrations.
Work with vector databases (Pinecone, Weaviate, Milvus, pgvector) and graph databases (Neo4j) to support knowledge management and intelligent retrieval.
Collaborate with frontend engineers to deliver conversational interfaces and AI-powered user experiences.
Optimize AI applications for scalability, reliability, and cost efficiency using cloud-native architectures on AWS or GCP.
Implement security best practices, authentication mechanisms, and prompt injection mitigation techniques.
Monitor and evaluate AI systems using model evaluation frameworks, RAG quality metrics, and LLM evaluation methodologies.
Troubleshoot and improve the entire AI lifecycle, from data ingestion through