Role:Specter is hiring a data operations engineer to build our research data operation. This individual will own the full pipeline from defining what data we need, to getting it labeled at high quality, to ensuring it meets the needs of our research team and ultimately improves our models. The role sits at the intersection of engineering and research, with a focus on building systems and tooling.
Responsibilities:- Own the end-to-end relationship with our data labeling provider, including task scoping, timeline management, and issue resolution
- Build and maintain internal tooling for labelers, including annotation interfaces, task pipelines, and dataset browsers
- Define and enforce quality control standards across all labeled data, implementing automated checks and audit workflows
- Partner with researchers to translate perception model needs into data collection strategies, identifying gaps in coverage across object types, scenes, lighting conditions, and sensor modalities
- Build dashboards and metrics to monitor dataset diversity, class balance, and domain coverage
- Close the loop on the data flywheel: track how labeled data flows into training, surface failure modes, and drive iteration on the pipeline from collection through to model improvement
- Evaluate and integrate new data sources
- Define labeling taxonomies and annotation specifications
Qualifications:- 1-3+ years of experience in data operations, project management, or a technical coordination role, ideally supporting ML or engineering teams
- Proficiency in Python and comfort building lightweight tools, scripts, and dashboards
- Strong written and verbal communication skills, with experience managing external vendors or cross-functional stakeholders
- Familiarity with ML workflows and how training data impacts model performance
- Highly organized, with a track record of managing multiple concurrent workstreams
- Self-directed and autonomous
- Bonus: experience with computer vision data, annotation platforms, or labeling operations
