Humanoid Manipulation Learning Ramp Leader

The mandate

Four weeks before an acquired robot-learning laboratory joins the group, the nominated integration leader withdrew because relocation was no longer possible. The legacy robotics team measures deterministic task completion, while the acquired scientists optimise policy learning from demonstrations; neither owns the decisions connecting data capture, skill promotion, safety constraints and repeatability at customer sites.

The incoming executive must establish a Tokyo base within four weeks and hold the combined seat for twenty months. A permanent global search opens after three priority skills clear controlled industrial pilots, expected in month fourteen, with an eight-week overlap planned for the successor to govern a learning portfolio review and a site-release decision.

Completion requires three manipulation skills to meet signed success, recovery and intervention corridors across ten representative site scenarios; every promoted policy must retain training-data, configuration and test lineage; the fleet-learning loop must survive an adverse-event drill; and the permanent appointee must approve the following two-quarter skill roadmap. A laboratory demonstration or curated video will not satisfy handover.

The interim may allocate robot hours and annotation capacity, pause policy promotion, set experiment priority, redeploy temporary specialists and contract simulation or teleoperation support below JPY 300 million. Hardware-platform changes, capital above JPY 1.5 billion, customer production releases and permanent executive appointments remain with the Robotics Investment Council; no schedule pressure can override a documented safety stop.

Locomotion architecture, actuator redesign, consumer-market planning and factory construction are excluded from the remit. The assignment owns the manipulation-learning system from demonstrations and synthetic variation through evaluation and controlled industrial transfer, while independent safety and application owners retain their statutory and customer accountabilities.

Why this seat is open

The acquisition closes against a fixed integration date, but the intended leader can no longer relocate. Allowing the two organisations to preserve incompatible evidence standards would postpone the difficult choices until customer pilots. A temporary executive is needed to integrate methods, talent and decision rights before the group selects the long-term owner of embodied learning.

What you will own

• Define the three priority manipulation skills as bounded operating problems with object, environment, human-presence, recovery and performance conditions.
• Reconcile demonstrations, teleoperation traces, simulation episodes and failed attempts into a versioned data ledger with consent, quality and coverage decisions.
• Establish promotion gates that connect offline policy evidence, hardware-in-loop trials, controlled-site variation, operator intervention and safe rollback.
• Decide how robot time, annotation effort, simulation capacity and application engineers are allocated across learning value and customer-readiness constraints.
• Integrate deterministic controls and learned policies through explicit arbitration, uncertainty handling, fallback behaviour and preserved incident telemetry.
• Command ten scenario-based pilot reviews, stopping deployment where object diversity, site variation or recovery evidence remains below the signed corridor.
• Hand the successor a functioning skill factory, experiment history, site exception map, partner commitments and witnessed quarterly allocation process.

Candidate qualifications

• Directed manipulation learning or embodied-AI industrialisation for humanoid, mobile-manipulation or advanced factory-robot systems beyond laboratory demonstration.
• Converted imitation, reinforcement or hybrid learning into repeatable physical skills across meaningful object and environment variation.
• Joined teleoperation, simulation, hardware tests and field telemetry into a traceable data flywheel with controlled policy promotion.
• Held deployment authority when learned behaviour underperformed and can show how safety evidence overruled a customer or investor timetable.
• Integrated acquired research talent with controls, hardware and application engineers without flattening materially different technical methods.
• Transferred a robot-learning organisation and decision system to a permanent executive after measurable industrial pilots rather than prototype showcases.

Non-negotiables

• Can relocate or establish a weekly Tokyo presence within four weeks and travel monthly to Nagoya, Osaka or another nominated pilot facility.
• Will treat the first twelve months as an exclusive executive engagement and participate in out-of-hours fleet escalation when required.
• Brings direct learned-manipulation deployment evidence; computer-vision leadership or conventional automation integration alone is insufficient.
• Must disclose current robotics investments, laboratory affiliations and commercial relationships with simulation, teleoperation or annotation providers.

1. 49 words maximum. Give your earliest Tokyo start date and identify any commitment incompatible with exclusive first-year leadership.
2. 49 words maximum. Describe one learned manipulation skill you moved into a variable industrial setting and the field evidence that delayed promotion.
3. 49 words maximum. Which data or control signal would make you stop a policy despite strong average task completion?