SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Gather AI is building a vision-powered platform using autonomous drones and existing equipment to digitize warehouse operations and supply chain workflows. The company combines computer vision, machine learning, robotics, and cloud infrastructure to deliver real-time warehouse intelligence.
As an ML Annotation QA Engineer, you will own the quality of annotated data across computer vision and machine learning programs. This is a judgment-heavy role focused on analysis that cannot be reliably outsourced. You will work with an external annotation partner at production volume, but the in-house responsibility is to analyze annotated data, make defensible quality calls the vendor cannot make consistently, and build aggregate views showing which facilities and equipment are degrading and why.
Key responsibilities include:
- Conduct daily review of annotated data and produce verdicts and root cause analysis in-house
- Own, version, and refine verdict taxonomy, decision rules, and quality guidelines
- Build and maintain performance trackers for error rates by facility, site, equipment, and data format
- Detect anomalies against baseline and flag them immediately
- Run root cause analysis to distinguish annotation error from model/system error from genuine field degradation
- Report findings to engineering and ML teams with reproducible evidence and stated confidence
- Identify systematic failure patterns and maintain a documented pattern library
- Query and analyze annotation data directly using Python and SQL
- Feed annotation-quality findings back as concrete SOP and instruction changes
- Specify annotation tool improvements and validate fixes
- Stand up quality analysis and reporting for new annotation programs
- Track work in Jira and contribute to pre-release validation
In your first 90 days, you will take over barcode and location root cause analysis from the annotation vendor, move from supervised review to owning the daily queue independently, and develop working expertise in the annotation pipeline and warehouse domain (racks, locations, bins, code formats, exceptions).
You will work closely with Machine Learning Engineers, QA, and Engineering teams. The first assignment focuses on warehouse forklift vision programs with barcode readability and localization analysis, expanding to drone imagery and new task types as customer capabilities come online. Success requires analytical rigor, sound judgment under ambiguity, and clear written communication.
