Rebar · posted · checked
Software Engineer, Applied AI
PythonPytorchTypeScript$150k – $250kHybrid · New YorkSenior
Who can apply
Hybrid · New York
Location: New York City
ABOUT REBAR
Construction is a $2 trillion industry in the US alone. Most of it still runs like it’s 1995: PDFs, spreadsheets, and email.
Every commercial building starts the same way. Someone opens a few hundred pages of blueprints and counts every piece of equipment, duct, and pipe by hand. Then they price it, bid it, and chase it for months. That work decides who builds what, and for how much. It has barely changed in 30 years.
Rebar is building the AI platform the trades run on. Our models read blueprints the way a veteran in the trades does. Our agents then do the work that follows: counting the job, pricing it, managing the bid, and tracking it through to the order. Counts that took three days now take one. Customers spend hours a day in Rebar, and thousands of new plan sets flow through it every week.
We started with HVAC contractors and their suppliers. Every other trade comes next.
Here’s where we go: when a building goes up, the trades that build it run on Rebar. Agents win the job, run the job, and buy the equipment.
We’re about 50 people in Union Square, growing fast, and backed by top investors. We’ve built only a fraction of what we need to.
ABOUT THE ENGINEERING ORG
The role of the engineer is rapidly changing. We're aware and we're being very intentional of ensuring we adapt with it. We are fostering an engineering culture of growth and development. We strongly emphasize care of craft and winning together. To echo our values, everyone operates like an owner, we find a way, and we win together.
ABOUT THE ROLE
This role is going to be sitting on our AI and ML team as a backend engineer.
You will be working directly with our in house models (as well as frontier VLMs) so that we can build a world class interaction for the frontend. For example, you could be taking a segmentation mask that one of our detection models produces, deciding what it should become on the canvas - a polygon, a polyline, a set of connection ports - designing the contract that carries it there, and then making it something our users can grab and correct.
The job is not just "make the model better", it’s to have the ideal shape and use case for the user can see, trust, correct, and bill against.
Show the full posting
RESPONSIBILITIES
Design the perception → product data contract