Engineering
Machine Learning Engineer, Vision
Turn a blurry photo of the inside of a fridge into a clean, correct ingredient list.
- Location
- Denver, CO, hybrid
- Type
- Full time
- Compensation
- $170,000 to $205,000 plus 0.3% to 0.7% equity
When someone snaps their shelf instead of typing, that photo has to become a list they barely need to correct. You own that.
This is applied vision with a product output. Real fridges are dark, crowded and half-labelled, and the bar is a list a tired person trusts at a glance.
What you would own
- Own ingredient recognition from photos: prompts, models, post-processing and the fallbacks when a shot is unreadable.
- Build the evaluation set of real kitchen photos and measure recognition against it every change.
- Decide when to ask the person to confirm rather than guess, because a wrong item is worse than a missing one.
- Keep the whole read inside one request and inside a second or two.
What we are looking for
- Experience shipping vision or multimodal features that real people used.
- You are comfortable working across model providers rather than betting on one.
- You measure before you tune, and you keep a held-out set honest.
- You want the credit for a feature that feels like magic and is mostly careful engineering.
How hiring runs
A thirty minute call with a founder, then a paid work session on a real problem from our backlog, three to four hours, then conversations with two people you would work with. Four steps, usually inside two weeks, and we tell you where you stand at every one.