AI Engineer
Build LLM, RAG and agentic systems that integrate with enterprise workflows, using rigorous evaluation, observability and responsible AI-assisted engineering.
Job Description
Mission Nexum is looking for an AI Engineer to design and build applications based on language models, retrieval, software agents and intelligent automation. You will move from prototype to reliable integration with explicit evaluation criteria, observability and controls. What you will do - Design LLM applications, RAG pipelines, tool-calling flows, multi-step agents and context-management strategies - Evaluate API-based and open-weight models against quality, latency, security and operating-cost requirements - Develop services and integrations using Python, TypeScript, FastAPI and enterprise APIs - Build retrieval and knowledge components with vector databases, embeddings and graph-based patterns where appropriate - Create repeatable evaluation datasets and measure groundedness, task completion, failure modes, safety and regressions - Implement logging, tracing, prompt and configuration versioning and runtime monitoring - Add guardrails, access controls and human-review points proportionate to each use case - Collaborate with product, data, platform and domain teams to integrate AI into real operating processes This job posting is addressed to both genders, in compliance with laws 903/77 and 125/91 on equal treatment in the workplace and against gender discrimination. We welcome candidates of all ages and nationalities, in accordance with legislative decrees 215/03 and 216/03. Nexum also encourages applications from people with disabilities, in compliance with current regulations.
How AI supports the role
Language models and agent frameworks are part of the systems you will build and also support coding, testing, documentation and evaluation. AI-assisted artifacts must retain provenance and be reviewed, tested and monitored before release. Sensitive data remains within approved environments and model limitations are made visible.
Tool choices vary by project. AI-assisted work remains subject to human review, security controls and documented validation.
What You Bring
LLM & Retrieval Systems
Model APIs or open-weight models, RAG, embeddings and vector databases.
Agent Orchestration
Tool calling, multi-step workflows, context management and agent frameworks.
Software Integration
Python or TypeScript, APIs, services, testing and enterprise integration.
Evaluation & Governance
Groundedness, regressions, observability, guardrails and human review.
How We Work
AI-Assisted Engineering
Approved AI tools accelerate analysis, implementation and documentation; people remain accountable for every released artifact.
Production Perspective
The work connects design, delivery, observability and continuous improvement rather than stopping at isolated prototypes.
Evidence & Traceability
Decisions are supported by tests, metrics, provenance and documented trade-offs.
Cross-Functional Delivery
Cloud, data, AI, security and business requirements are connected around real operating constraints.
