PSenior AI Engineer /ppAptus.AIstrong · Pisa, Italy (Hybrid) /strong /ppbrTutti i candidati sono invitati a leggere attentamente la seguente descrizione del lavoro e le relative informazioni prima di candidarsi.br/ppAbout Aptus.AI /ppAptus.AI is building the next generation of AI for the legal and compliance industry. Our platform combines frontier language models with deep domain expertise to help legal professionals and compliance teams work faster, more accurately, and with greater confidence. We operate at the intersection of cutting-edge AI research and the rigorous demands of the legal world, and we're growing fast. /ppbr/ppAbout the Role /ppWe're looking for an AI Engineer who has already built real AI agents, not just integrated an LLM API, but designed and shipped systems that reason, plan, and act autonomously. You'll work at the core of our product, designing the multi-agent architectures that power our compliance and legal AI assistants. /ppThis role sits at the intersection of systems design, applied AI, and product engineering. You'll collaborate directly with our product and engineering leads to translate complex legal workflows into robust agentic pipelines, and you'll own those systems end to end, from architecture to production. /ppWe're looking for someone who thrives in ambiguous, fast-moving environments, picks up new technical domains quickly, and holds themselves to a high bar for code quality and system reliability. /ppbr/ppResponsibilities /pulliDesign and implement multi-agent systems for legal and compliance use cases, including orchestration patterns, tool use, memory, and state management /liliOwn agentic pipelines end to end: from requirements and architecture through deployment and iteration /liliWork closely with product, legal domain experts, and customers to understand workflows and translate them into technical solutions /liliPrototype quickly, gather real-world feedback, and iterate with discipline /liliContribute to architectural decisions and help establish engineering best practices for AI systems /liliCommunicate clearly with both technical teammates and non-technical stakeholders about system behavior, trade-offs, and progress /li /ulpbr/ppYou may be a good fit if you /pulliHave already designed and shipped AI agents in production, systems that plan, use tools, and operate with meaningful autonomy /liliHave a solid grasp of agent orchestration patterns (ReAct, multi-agent delegation, tool-calling loops, routing, parallelism, etc.) /liliWrite clean, maintainable code and have strong software engineering fundamentals /liliCan quickly develop deep understanding of complex, unfamiliar technical or legal domains /liliAre proficient with modern web technologies (e.g. Python, TypeScript) and comfortable working across the stack /liliCommunicate effectively with both technical and non-technical stakeholders, you can explain what an agent is doing and why to a lawyer as clearly as to an engineer /liliHave a high tolerance for ambiguity and a bias toward action /li /ulpbr/ppStrong candidates may also have /pulliExperience with LLM orchestration frameworks (LangGraph, LlamaIndex, CrewAI, custom implementations, etc.) /liliFamiliarity with RAG architectures and vector databases (e.g. Pinecone, Weaviate) /liliBackground in legal, compliance, or other regulated domains /liliExperience building internal tooling or developer-facing platforms /liliExposure to streaming architectures and real-time UI patterns for LLM output /liliA history of iterating directly with end users and incorporating feedback into system design /liliHands-on experience with AWS services relevant to AI workloads (e.g. S3, Lambda, ECS/EKS, SageMaker, IAM) and comfort navigating cloud infrastructure independently /li /ulpbr/ppRepresentative Projects /ppHere's the kind of work you'd own at Aptus.AI: /pulliDesigning a multi-agent pipeline that routes legal queries to specialized sub-agents (case law retrieval, regulatory analysis, document drafting) and synthesizes a coherent response /liliBuilding an AI compliance assistant grounded in a proprietary risk matrix, capable of answering structured and open-ended questions about corporate liability /liliCreating the orchestration layer that decides whether a query needs fast single-step reasoning or a deeper, iterative research workflow /liliDeveloping full-stack tooling that lets non-technical legal experts inspect, evaluate, and give feedback on AI agent outputs /li /ulpbr/ppWhy Aptus. xdwybme AI /pulliLegal AI is one of the most consequential domains where AI can have real impact, the work matters /liliYou'll have high ownership and direct influence on product and architecture from day one /liliWe're based in Pisa with a highly-flexible hybrid setup /li /ulpbr/ppWe encourage applications from candidates who don't meet every single requirement. If you're excited about the problem and bring strong fundamentals, we want to hear from you. /ppbr/p