Who we are
Nivara is an enterprise AI company based in Turin. We build systems that take artificial intelligence out of the demonstration and into the working day of a large organisation — using its own data, within its own constraints, and under its own governance. We are a small team, growing quickly, working on engagements where AI is not a roadmap feature but the thing the business runs on. Our name comes from a line we mean: the shelter of reason.
Why this role exists
Most enterprise AI pilots never deliver measurable results. The gap is not the model. The gap is production — and closing it requires someone to be physically present. That is this job.
One position. Immediate start. If you are the right person, we will move in days, not months
This is a single hire, not a pipeline. The seat is needed now. We review applications as they arrive rather than in batches; we interview immediately, and we do not delay decisions. When we make an offer, we ask for an answer within 48 hours — we are telling you this now, in the advertisement, so it is never a surprise at the end. The position closes as soon as we find the right person, which could be next week. If this is your job, apply today.
Two things you need before you apply — please read them. Fluent English and Italian are essential
You will spend your days inside Italian client organisations, in their operating meetings, with the people who actually run the process being automated. English is the working language of our team; it is not the working language of the room you will be in.
You must already hold the right to work in Italy, and be able to start on that basis now
At this stage, we are not in a position to sponsor visas or support any immigration procedure — no applications, no renewals, no conversions, no nulla osta. If your ability to work in Italy depends on a procedure that is still open, this is not the right moment for us, and it is fairer to state this at the top of the advertisement rather than at the offer stage.
Where you work, plainly stated
You work within our clients' organisations. You attend the operating meetings for the process being automated, see real data and edge cases, and carry the first use case from demonstration into production. This means regular travel to client sites in Italy, as an ordinary part of the role. Between engagements, your base is Turin: three days on site and two remote, under an individual agile-working agreement — and until our Turin office opens, on-site days are suspended and you work fully remotely.
What you own
You take the first use case from demonstration to production, and you own its reliability once it is live — you are measured by what happens in service, not by what shipped. Then you transfer the method and capability to the client's own engineers, so the second use case costs them less than the first. What you learn in production feeds back into our platform and our standards.
The balance
Roughly four out of every five days are spent on post-contract implementation, hardening and stabilisation. The remaining share is technical scoping and demonstrations for engagements already in advanced discussion. We are telling you the split because getting it wrong is the most common way this role disappoints the person doing it. You will run about one and a half concurrent engagements — deliberately.
The stack, specifically
Python as the working language, with Go or TypeScript welcome alongside. REST and gRPC APIs, distributed systems. PostgreSQL, Clickhouse, Elastic, Redis; vector stores — Milvus, Pinecone, pgvector. Frontier and open-weight models via their APIs: OpenAI, Anthropic, Gemini, Llama. Orchestration and evaluation with LangChain, LlamaIndex or DSPy, and an evaluation harness you actually run. Docker, Kubernetes, Terraform, CI/CD. AWS, Azure or GCP. MLflow or Databricks on the MLOps side.
Skills
Python · Large Language Models (LLM) · Retrieval-Augmented Generation (RAG) · Kubernetes · Terraform · PostgreSQL · MLOps · Italian · Amazon Web Services (AWS) · Client-Facing Delivery
Compliance is engineering here
For regulated processes, the EU AI Act and the GDPR apply to the same system at the same time. You build data minimisation, pseudonymisation before inference, role-based access control and traceable logging as technical controls — at design time, not as a document written afterwards.
What we will not compromise on
Production experience with LLM applications, agentic systems, or workflow orchestration — systems that handled real traffic and had someone accountable for uptime; notebooks and internal pilots are not this. Real backend engineering. End-to-end ownership of client-facing delivery. And the ability to work in ambiguity and make it smaller, not louder.
Terms
Permanent employment under Italian law (contratto a tempo indeterminato), CCNL Terziario, Distribuzione e Servizi, Level I. Six-month probation, as set out in the collective agreement for this level. The package comprises three parts — contractual gross annual salary, company welfare, and eligibility for the group's financial participation plan in force from time to time — and we set out all three at first contact. A company welfare plan open to the whole workforce is being introduced: supplementary healthcare, pension top-ups, training, administered through a welfare platform.
Start date. As soon as you can — the seat is open now
In your application, please state your earliest possible start date and, if applicable, your notice period in days. If you are available immediately, state this on the first line: we will move at that speed.
Apply via LinkedIn today!