Neuromorphic Event-Sensor Bring-Up Leader — Warehouse Robotics

The mandate

The sensor-platform head role requires leading perception teams working on event-sensor integration for warehouse robotics, addressing challenges including hot-pixel bursts, timestamp drift and bias shifts under LED flicker, reflective wrapping and rapid temperature change. The leader must deliver sensor-to-robot evidence without allowing software suppression to hide unstable silicon, optics or timing behaviour, ensuring benchmark accuracy, consistent braking latency and small-object detection across sensor lots.

Eight months provide a fixed bring-up window. The first month secures device and board genealogy, reproduces factory conditions and establishes raw-event baselines. Months two through five isolate pixel, bias, clock, optics and firmware mechanisms and qualify production screens. Months six and seven prove mixed-lot robotic performance. The final month installs a permanent leader and transfers safety, supplier and volume-release decisions.

Handover is achieved when event-rate distributions, timestamp accuracy and detection latency meet approved limits across three production lots and representative warehouse lighting for eight consecutive weeks. The successor must chair one sensor-lot disposition, approve a bias or filter change and defend its braking consequence to product safety. Hidden event clipping, laboratory-only illumination and hand-tuned per-device settings cannot count as stable readiness.

The leader may quarantine sensor lots, stop robot trials, approve bias and firmware parameters within safety limits, allocate failure-analysis capacity and commit EUR 3 million within the authorised recovery envelope. They decide sensor readiness for controlled pilots after safety consultation. Robot motion policy, site deployment, product safety acceptance, supplier awards and customer commercial remedies remain with named accountable executives.

Excluded are redesigning the warehouse robot, owning perception-model research, changing safety goals, conducting uncontrolled live-site experiments or replacing the full sensor architecture. The mandate covers event-sensor bring-up, integration evidence and production control for the selected platform. Patent attribution, supplier liability and employee conduct are handled outside the technical recovery.

Why this seat is open

Event cameras expose temporal information conventional frames miss, but their output is highly sensitive to device bias, timing, illumination and downstream filtering. The departure left silicon and robotics teams able to optimise separate benchmarks while no executive owns safe end-to-end latency. Temporary leadership can restore raw evidence now and leave a successor who can balance device stability with perception performance.

What you will own

• Reconcile sensor lot, pixel map, optics, board, clock, bias, firmware, temperature and robot-test genealogy.
• Separate hot pixels, background activity, flicker response, timestamp error, bandwidth saturation and perception-pipeline artefacts.
• Define raw-event and task-level measures connecting device behaviour to detection, tracking, braking latency and false stops.
• Govern environmental testing across illumination spectra, reflectivity, motion, vibration, temperature and electromagnetic conditions.
• Stabilise bias, calibration, filtering and bandwidth settings without erasing safety-relevant events or creating device-specific tuning.
• Direct supplier containment, production screening, firmware correction, perception compensation and lot-release evidence.
• Transfer failure signatures, safe parameter envelopes, production tests, field monitors and successor-led pilot decisions.

Candidate qualifications

• Led event-based vision, imaging-sensor or neuromorphic silicon bring-up into a production robotic or automotive system.
• Can evidence a software filter that improved benchmark accuracy while concealing unsafe sensor instability or latency.
• Understands pixel behaviour, event thresholds, timestamping, optics, lighting flicker, bandwidth and perception integration.
• Has governed physical robot trials where incorrect sensor conclusions could affect people or equipment.
• Directed semiconductor suppliers and algorithms teams through shared raw-data, genealogy and mechanism evidence.
• Developed a successor able to reject attractive perception metrics when device-level controls remain unstable.