Senior Software Engineer, Architecture

About Navflex

At Navflex, we're pioneering the future of logistics automation through cutting‑edge AI and robotics. Our autonomous mobile robots (AMRs) are transforming the way goods are loaded and unloaded, enabling plug‑and‑play solutions that streamline operations and enhance efficiency across the global supply chain for some of the world’s most demanding warehouse Environments. Our international, cross‑disciplinary team in the EU and USA pairs robust mechatronics with cutting‑edge navigation and perception to deliver safe, reliable autonomy in real‑world warehouses and yards.

Join us in shaping the future of intelligent logistics.

You'll own the application and integration layer of Navflex's autonomous forklifts — the software that turns a transport order from a customer's WMS or fleet manager into a safe, observable, repeatable sequence of robot behavior, plus the infrastructure that ships that software onto vehicles in the field.

This is the direct counterpart to our Algorithms group: where they own perception, localization, and planning, you own the system those algorithms run within — task/order state machines, external interfaces (VDA
5050/MQTT, LIF, geoJSON, WMS, Ether Net/IP, CAN), telemetry, and the real-time execution model that allocates our heterogeneous multi‑core industrial PCs across critical control loops. You’ll be the real‑time systems performance engineer specializing in software diagnostics, low‑latency tuning, and advanced bottleneck analysis.

You're part of the team responsible for the distance between “it worked on the test robot” and “it works on 70 robots across four sites” — and, when it doesn't work, for making sure we know why within minutes.

• Latency budgets are documented, measured, and met on real target hardware — control-loop timing is a number we can show a customer, not an anecdote, and it holds as the stack grows.

• Time-to-root-cause for field issues drops from days to minutes — on-vehicle telemetry and traces answer "why did this robot stop?" without a site visit or a reproduction attempt.

• Releases reach the fleet predictably — reproducible builds, safe rollout and rollback across sites

• New customer integrations land on schedule — bringing up an unfamiliar WMS or fleet manager becomes a known‑effort task rather than a research project

• The architecture absorbs growth — new behaviors and new sites ship without cross‑cutting rework, and component boundaries hold as the fleet scales.

• Robot application architecture - task/order state machines, workflow orchestration over the navigation stack, exception and recovery handling, and boundaries that keep algorithms, hardware I/O, and business logic independently testable and replayable.

• Real-time execution & compute budget - CPU pool design, core pinning and isolation, scheduling policy, and the single‑source‑of‑truth allocation strategy that applies to launch files and containers.

• Performance debugging - root‑cause jitter, missed deadlines, and contention across the stack (CPU/cache/interrupt pressure, middleware overhead) using our RT Linux tooling (cyclic test, Perfetto, per-node profiling, and similar).

• External integrations - build and harden VDA
  5050/MQTT, LIF, geoJSON, WMS flows, and industrial I/O (Ether Net/IP, CANopen), resilient to unreliable networks and non‑conforming counter parties.

• On‑vehicle data & telemetry - persistence layer for order history, metrics, and debug traces, with versioned migrations that survive field upgrades and never compete with control for cycles.

• Standards & mentorship - participate in testing strategy design (unit, simulation, replay‑based regression), and code review.

• Experience contributing to and maintaining…