The Opportunity
Manufacturing powers the global economy at $50T a year and it relies heavily on human dexterity and skill to produce the goods we rely on every day. Yet manufacturers have little visibility when issues arise at manual assembly stations, impacting productivity and quality.
Assembler AI is changing that.
We use computer vision and artificial intelligence to help manufacturers improve quality, reduce waste, increase throughput, and enable frontline employees to perform at their best.
Backed by Diagram Ventures, we're building a category-defining company at the intersection of AI, manufacturing, and operational excellence.
The role
We're hiring our first Data Annotator. You'll work directly with the AI team on the labeled video that every one of our models is trained and verified on, and take ownership of labeling quality across our customers' stations.
This is a hands-on, detail-driven role. Our models are only as good as the judgment behind the labels: whether a step really started on that frame, whether a flagged event actually happened, whether a label scheme still fits a new station. If you're the person who notices the one clip out of four hundred that doesn't match the others, you'll do well here.
What you'll do
- Review video clips flagged by our models and decide whether the event is real, using our keyboard-driven review tool
- Segment work steps in station recordings and draw bounding boxes on video frames
- Correct model output: adjust event timing, fix labels, and document what the model got wrong and why
- Maintain labeling guidelines and keep them current as new stations and customers come online
- Run calibration checks and quality reviews so labels stay consistent over time and across people
- Feed back on the tools and the process, and help onboard and train future annotators
Requirements
- A sharp eye and the patience to stay consistent across hundreds of near-identical clips a day
- Comfortable making a call on ambiguous footage, and saying "not sure" when it is
- Curious about why a model fails; you notice patterns, not just single mistakes
- Organized: you keep guidelines, edge cases and decisions written down
- Able to work remote 5 days a week and come into the office once a month
Technical environment
Required:
- Comfortable working all day in browser-based tools with keyboard shortcuts
- Basic spreadsheet skills for tracking throughput and quality
Nice to have:
- Prior data labeling or annotation experience, any modality
- Manufacturing, quality control, or factory floor experience
- Familiarity with annotation platforms (Label Studio, CVAT, Roboflow) or basic Python
- Interest in how machine learning models are trained and evaluated
Why join
You'll be the first person dedicated to data quality, with direct ownership of the labels that decide whether a model works, and a very short path from your work to something running on a live production line. The role grows with the company: into annotation lead, QA, or model evaluation, depending on where you want to take it.
Compensation & Benefits
- Competitive salary
- Competitive health benefits
- Health and wellness spending account, from $500 to $1,000 annually
- Latest MacBook
- Opportunity to grow as the company scales