Data Labeling Ops Lead: Build Trusted Datasets

Bobyard; San Francisco (CA); On-site; USD 90,000 - 125,000; Full time

Benefits offered by this job

Equity

Job summary

Bobyard in San Francisco is building AI to bring visual intelligence to construction. You will own the labeling operation end to end, leading the annotator team, setting quality bars, and delivering datasets that ML engineers can trust and rely on.

This is a first-in-function hire—no playbook. You’ll recruit and train annotators, improve labeling workflows, clean up existing data, source new drawings, scope ML requests, and partner with engineers to fix model gaps with better data.

Qualifications

• Direct experience managing a labeling, annotation, or data-quality team.
• Extremely detail-oriented — you notice inconsistencies before anyone points them out.
• Strong operational instincts — can run many datasets and priorities without dropping details.
• Technical enough to work with ML engineers and understand false positives/negatives, class imbalance, train/test splits.
• Resourceful — you find new data when needed.
• High ownership — ensure datasets are actually good and shipped.

Responsibilities

• Build and lead the annotator team: recruit, onboard, train, and uphold quality and throughput.
• Own labeling quality: review annotations, catch errors, and refine guidelines.
• Clean up existing datasets: fix labels, metadata, duplicates, and other issues.
• Source new data: organize construction drawings to expand formats and edge cases.
• Turn ML requests into shipped datasets: scope, run, and deliver clean data on time.
• Collaborate with ML engineers to address model failures with better data.
• Build tools/workflows to speed labeling and reliability.

Skills

Team management

Attention to detail

Operations

ML collaboration

Problem solving

Ownership

Tools

SQL

Python

Job description

Bobyard in San Francisco is building AI to bring visual intelligence to construction. You will own the labeling operation end to end, leading the annotator team, setting quality bars, and delivering datasets that ML engineers can trust and rely on.

This is a first-in-function hire—no playbook. You’ll recruit and train annotators, improve labeling workflows, clean up existing data, source new drawings, scope ML requests, and partner with engineers to fix model gaps with better data.