Technisch engineer.
Pubblicato il 01-10-2026 - Lexroom in Italia
At Lexroom.ai, we set the bar for legal AI—clear, fast, and built for trust. In just a few years, what began as an idea has become a reliable platform adopted by thousands of legal professionals across Europe. Our mission is clear: make the law faster, clearer, and more reliable, giving lawyers, law firms, and companies the time and confidence to focus on what truly matters. Today, Lexroom is a trusted partner, shaping the evolution of legal AI with the same rigor and precision that define our product. Why Lexroom.ai?
Shape the Standard: Be part of the team setting the benchmark for legal AI in Europe.
Your contribution will empower legal professionals to make faster, clearer, and more reliable decisions worldwide. At Lexroom.Here, you'll build something that lasts: a trusted platform, a strong reputation, a new standard for the legal world.
Design and build agent-driven and RAG-based AI systems to solve complex legal tasks with precision and efficiency.
Fine‐tune and post‐train LLMs using cutting‐edge supervised and reinforcement learning techniques to raise response quality.
Develop and optimize large‐scale search pipelines for LLM‐powered agents.
Run end‐to‐end evaluations to pinpoint bottlenecks, measure performance, and guide continuous improvements.
Maintain technical excellence by following established best practices, coding standards, and rigorous code‐review processes. Proven expertise in building and managing robust ML pipelines.
Strong Python skills with hands‐on experience in LLM research and deep learning systems.
Knowledge of search technologies such as full‐text search, vector databases, or other large‐scale retrieval systems.
Collaborative mindset and readiness to work with legal experts, researchers, and engineers to keep improving the product.
Familiarity with MLOps and integrating ML models into production backends (a strong plus).
Up‐to‐date with LLM innovation, ideally with experience in pre‐training or deploying large‐scale models in real‐world