Annotation Operations Manager Redwood City, CA Fulltime

Dyna Robotics trains robots to do real world manipulation tasks, and every one of those behaviors is learned from precisely labeled robot episodes. Our Data Annotation team turns raw teleoperation footage into the labels our imitation learning models depend on, so throughput and quality here are a direct input to how fast the company ships.

Dyna Robotics has raised over $140M, backed by top investors including CRV, First Round Capital, Robostrategy, Salesforce Ventures, NVentures, Amazon, Samsung Next, and LG Technology Ventures. Our team brings together engineers and researchers from Google, Meta, Apple, Amazon, Cruise, Aurora, NVIDIA, along with academic roots at Stanford, Berkeley, MIT, UPenn, and beyond. We’re positioned to redefine the landscape of robotic automation.

This role owns the operational layer of that work. You will run the operation from day one, covering process, reporting, and third party relationships, and step into people management progressively, starting with a small pod of labelers and growing into the full team as you prove yourself. Our Annotation Lead keeps the technical judgment calls: model calibration and promotion, ontology and failure taxonomy design, and cost and planning strategy.

You are the person who makes the annotation engine run predictably every single day.

Operations and reporting

• Own the reporting rhythm. Run daily and weekly throughput reporting across all active datasets, covering episode totals, review stage progress, and ETAs, and distribute scorecards to the labeling team and stakeholders.

• Keep the pipeline moving. Own the Encord pipeline end to end: create and configure projects, set up SOPs and ontologies, move datasets through annotate, review, and complete stages, handle grade splits, assignments, and resyncs, and keep dataset updates flowing as new footage lands.

• Measure and unblock. Track and continuously improve operational metrics such as throughput per headcount, per dataset cycle times, and expected versus actual labeling time, and flag bottlenecks before they turn into blockers.

• Automate the mechanical. Extend and maintain the automation and runbooks behind these tasks, like the scripted daily totals send, so manual toil shrinks over time.

People and team management

• Start with a pod, then scale up. You will begin by leading a small pod of labelers, owning their day to day assignments and first line supervision. As that proves out, you will transition into managing the full annotation team, about 12 people today and growing, and formalize sub leads as it scales.

• Handle approvals. Own timecard, payroll hours, and approval workflows for the team.

• Grow the team. Run hiring for the annotation team end to end, including sourcing, interviewing, onboarding, and ramp.

• Develop people. Own performance management: regular 1:1s, performance reviews, coaching, and quality feedback loops.

Third party and vendor management

• Own the relationships. Be the primary point of contact for our annotation platform (Encord) and for external labeling vendors and partners.

• Coordinate partners. Run external work streams by provisioning access and invites, standing up projects and SOPs for partners, and managing scope and priorities with them.

• Manage the handoffs. Own the scale and vendor upload runbook, and make sure data moves cleanly between our systems and external tooling.

• Absorb the overhead. Represent annotation ops in vendor and external syncs so the rest of the team stays focused.

• 3+ years running operations, program management, or team management, ideally in data annotation, data operations, ML data pipelines, BPO or vendor management, or a comparable high throughput environment.

• Direct people management experience. You have supervised a team, run hiring, and handled performance and approvals.

• Strong operational instincts. You build repeatable processes, track the right metrics, and are comfortable owning dashboards, scorecards, and reporting.

• Comfort with annotation and labeling tooling such as Encord, plus enough technical fluency to work in spreadsheets, basic SQL or scripts, and pipeline tools without hand holding.

• Vendor and third…