Software Engineer

3 ore fa

bari, puglia, Italia Mirai s.p.a. Tempo pieno

Mirai is a European deep tech company building Physical AI for distributed autonomy. Mirai OS lets robots understand context, take initiative and sustain operations together — one autonomy layer deployed across Mirai Frames and third-party platforms. We build for defence, government and dual-use customers, maritime first.

Our autonomy stack works as one system: the software running on the vessel, the Rust engines that recreate it in a digital twin, the behavioural logic that decides what it does next, the communication layer that keeps it connected, and the AI that reasons on board — explainably, and within the limits we set. We need senior engineers who can move across all of it.

This is a hands-on role. You will bring a solid sense of architecture, but your job is to write code that works — in real time, on real hardware, at sea. Rust is how we build the core of our stack, so experience with it is a strong plus.

We are looking for the kind of engineer who treats AI-assisted development as a core competency — Cursor, Copilot, Claude, whatever gets the job done — and uses it to ship robust, production-grade software faster than anyone around them.

We're hiring several engineers for this role, and we expect each of them to work across the whole architecture rather than in a single corner of it.

Key ResponsibilitiesAutonomy & Real-Time Systems
  • Develop and integrate the software of Mirai's autonomous system, from onboard real-time components to the services that coordinate vessels and missions.
  • Write deterministic, low-latency code that holds up in real operating conditions — degraded links, constrained compute, no human fallback.
Digital Twin & Simulation
  • Build and extend the Rust engines behind our digital twin, so vessels, sensors and environments can be simulated faithfully before anything touches the water.
  • Close the loop between simulation and field data, so what we test virtually matches what happens at sea.
Algorithms & Behaviour
  • Design and implement the algorithms and behavioural engines that turn mission goals into vessel decisions.
  • Develop the planning and collision-avoidance logic that keeps vessels safe and compliant at sea.
Edge AI & Reasoning
  • Integrate reasoning models that run on the edge, so each vessel can interpret context and mission intent and act within the authority it has been given.
  • Fine-tune and optimise vision pipelines and other models for our sensors, our domain and constrained edge hardware.
  • Make AI decisions explainable and bounded, together with our safety engineers: guardrails, decision boundaries and escalation to a human operator, with every decision traceable.
Communication & Integration
  • Build and maintain the event-driven communication layer between vessels, ground systems and services — NATS JetStream at the core, MQTT for interoperability with external systems.
  • Own the interfaces between components, so onboard software, simulation and cloud behave as one system.
QualificationsRequired
  • Master's degree (Laurea Magistrale) in Computer Science, Computer Engineering or Software Engineering.
  • 4+ years of professional software engineering experience.
  • Production-level C++ or Rust, plus Python.
  • Proven experience developing algorithms that run on real-time systems, with a clear grasp of concurrency, timing and determinism.
  • Strong mathematical foundations — linear algebra, probability and estimation, geometry.
  • Experience with event-driven distributed systems: pub/sub, event logs, replay and delivery guarantees.
  • Testing discipline for real-time code: simulation-based testing, software- and hardware-in-the-loop, CI.
  • Solid software architecture skills, applied pragmatically: you design so that the code you write works, ships and keeps working.
  • Practical experience applying AI/ML in real systems — beyond notebooks.
  • Proficiency with AI-assisted development tools as part of your daily workflow.
  • Fluent English.
  • EU or NATO-country passport with unrestricted right to work in Europe; eligibility for a national security clearance.
Preferred
  • Production experience in Rust — the language behind our digital twin and the core of our stack.
  • ROS 2 in production, real-time control, and path and motion planning — including collision avoidance under the COLREGs.
  • Behaviour trees, decision-making algorithms or reinforcement learning applied to control.
  • Sensor fusion and multi-target tracking across radar, LiDAR, EO/IR cameras, AIS and GNSS/INS.
  • Running, optimising and fine-tuning models on edge hardware such as NVIDIA Jetson — vision pipelines and reasoning models (LLMs or VLMs) alike.
  • Explainable AI and guardrailing: bounding what a model m