Robotics Data Collection Delivery Manager

Innodata Inc.

9d•$120,000 - $180,000

About The Position

The next frontier for AI is the physical world, and Innodata is building the data factory that powers Physical AI for frontier robotics and humanoid systems. As a Robotics Data Collection Delivery Manager, you will own end-to-end delivery of egocentric video data collection programs for some of the world's most innovative robotics clients — combining deep operational execution with a working knowledge of robotics architectures, data pipelines, and physical AI systems. This is a high-visibility role that directly influences strategic client relationships and Innodata's $100M practice opportunity in egocentric robotics data. This is a ground-floor opportunity to build and lead a new practice area at Innodata at a moment when demand for egocentric robotics training data is accelerating rapidly. The person in this role will sit at the intersection of physical AI, large-scale data operations, and frontier robotics — with direct impact on a $100M practice opportunity and working relationships with some of the most innovative robotics companies in the world.

Requirements

A completed Bachelor's degree, or higher, in a related field
5+ years of experience in data collection, content operations, or program management, with at least 2 years managing large-scale contributor or workforce programs
Hands-on familiarity with robotics or Physical AI systems — including perception/control stacks, egocentric data pipelines, teleoperation, or simulation — strongly preferred
Working knowledge of robotics model architectures (VLMs, VLAs, ACTs, world models) and how training data requirements flow from model needs
Experience with egocentric video data collection, computer vision datasets, or multimodal robotics training data
Strong operational and project management skills — comfortable managing multiple concurrent workstreams across geographies with aggressive timelines
Proven ability to translate complex technical specifications into clear, actionable guidance for non-technical contributors
Experience working with clients in a B2B delivery context, with excellent communication skills across both senior executives and front-line engineers
Comfort working cross-functionally with Sales, Product, Engineering, and Operations in a fast-moving, ambiguous environment

Nice To Haves

Prior experience in data/AI/robotics delivery, solutions engineering, or technical program management a strong plus

Responsibilities

Own end-to-end delivery of egocentric robotics data collection programs from kickoff through final hour acceptance, ensuring all client quality, diversity, and timeline targets are met
Develop and maintain a standard collection playbook adaptable across clients with adjustable parameters — camera geometry, naturalism blend, mount type, regions — rather than rebuilt from scratch each engagement
Build reusable delivery assets including SOPs, onboarding and training curricula, scorecards, rubrics, and tracking infrastructure shared across client programs
Monitor daily project progress against hour targets and acceptance rates, identifying and resolving bottlenecks before they impact delivery
Act on client QA feedback within 24 hours, relay actionable guidance to contributors, and drive measurable improvement across quality, diversity, and task adherence dimensions
Collaborate closely with Robotics Solutions Engineers to ensure clean handoff from pre-sales design to delivery, aligning on scope, success metrics, and data collection execution plans
Develop a working understanding of client robotics architectures — including VLMs, VLAs, ACTs, world models, and manipulation policies — to translate model requirements into concrete data collection specifications
Translate client technical requirements (camera geometry, IMU specifications, naturalism targets, egocentric capture pipeline standards, file format requirements) into clear contributor guidance and internal delivery standards
Partner with clients’ robotics and ML teams to understand evolving data needs and proactively adjust collection programs to ensure dataset utility for model training
Maintain fluency in egocentric capture best practices, sensor/rig configurations, and common failure modes (e.g. robotic movement, poor framing, low task diversity) to engineer quality upstream
Oversee a continuous contributor recruitment funnel capable of sourcing and onboarding 10,000+ participants at scale across US, LatAm, and developed Asia markets
Partner with recruiting channels — crowd work agencies, BPO and workplace partnerships, universities, gig platforms, and referral networks — to maintain a reliable, diversified sourcing pipeline
Design and enforce participant onboarding and training programs that front-load quality, addressing common failure modes before volume collection begins
Monitor contributor-level metrics including acceptance rates, fraud rates, diversity scores, and hour contribution, taking corrective action where needed
Serve as the primary operational point of contact for robotics clients during active collection engagements, maintaining strong relationships and positioning Innodata for contract expansion
Provide regular project status updates to clients and internal leadership, including hour delivery progress, acceptance rates, fraud rates, and open issues
Bring structured, frontline feedback from clients back into Innodata’s product and roadmap discussions, helping shape a scalable Robotics & Physical AI data practice
Represent Innodata’s robotics data capabilities at key industry events and contribute to thought leadership through talks and published artifacts
Coordinate device procurement, configuration, shipping, and recovery across collection regions, with a preference for company-owned, centrally managed iPhone fleets over participant-owned devices
Standardize on client-defined hardware rigs — head mount, tilted-down geometry — and mitigate known comfort issues via mount fit, session length guidance, and participant training
Build and manage a logistics function to procure, configure, track, refurbish, and redeploy devices across LatAm, Asia, and US markets
Enforce fraud prevention protocols across the contributor base, ensuring confirmed fraudulent sessions are flagged, not compensated, and that fraud rates stay below client thresholds
Drive a quality-first collection culture through training, real-time monitoring, and fast feedback relay — treating QA as an upstream engineering problem, not a downstream fix
Where clients provide backend QA access, plug into the feedback loop to continuously refine training and delivery