At Pillar, we are building the financial operating system for construction.

Construction is one of the largest industries in the world. It moves trillions every year, yet it still runs on slow, fragmented processes and outdated tools.

We are changing that with AI and a fast, simple, scalable product designed for the companies that actually build our cities.

Pillar is an Italian startup that recently raised one of the largest pre-seed rounds in Italy, and we are building an international company for the construction industry.

We are moving fast. If you want to write code that ends up in the hands of the people building our cities, you’re in the right place.

We are hiring a Full Stack Engineer to join our compact team of A+ people and help build the foundations of Pillar.

• Design and develop agentic workflows: prompt engineering, tool orchestration, multi-step reasoning, and production deployment
• Build RAG and semantic search pipelines using vector databases and embedding models to provide relevant context to AI agents
• Develop document extraction pipelines that transform unstructured files (invoices, reports, forms) into structured database records
• Set up observability and evaluation infrastructure: tracing, experiment tracking, and automatic scoring of LLM output quality
• Optimize context strategies for agents: chunking, hybrid search, re-ranking to improve response quality
• Work across the entire stack (React / Next.js, tRPC, PostgreSQL, serverless functions) when AI features require it

• Experience building LLM-powered applications in production: agents, tool use, structured outputs
• Hands-on experience with RAG systems: embeddings, vector databases, hybrid search, chunking strategies
• Experience with TypeScript, or strong interest in learning it (our stack is TypeScript end-to-end)
• Experience with AI orchestration frameworks (LangChain, Mastra, CrewAI, or similar)
• Strong knowledge of SQL and relational databases: you’ll write queries and optimize data pipelines
• Understanding of prompt engineering techniques: few-shot prompting, chain-of-thought, structured output parsing

• Experience in MLOps: deploying models on serverless GPU platforms, CI/CD for AI systems, monitoring agents in production
• Experience with MCP (Model Context Protocol) or similar standards for tool distribution
• Background in multi-turn conversational agents with memory and state management
• Familiarity with event-driven architectures and message queues for asynchronous AI workloads

• You prefer working alone: successes and failures are shared, and collaboration is essential
• You need rigid structure: Pillar embraces chaos and operates with a fluid, fast-moving organization
• You don’t take ownership: teams at Pillar own their work end-to-end, and initiatives often start bottom-up

• Competitive salary + stock option plan: share the value you help create
• You’ll come to the office for onboarding, company events, or whenever you want
• A small, ambitious team: zero bureaucracy, maximum impact