Struction is different

Reading through all these job descriptions suck. They all say the same s*** in tech lingo to signal their access to the most advanced models at the coolest office with the best incentive package. You probably applied to a few “mature” series A+ companies freshly funded by A16z or Sequoia, or maybe Big Tech as a Sr CV Engineer. We all know how those jobs turn out. You get a sit/stand desk and play baby sitter for the mercenary CTO or PM on Manager Mode for +10 hours per day (cringe). You try to play the politics game to raise more coin until you realize the futility. You hit your 1 year cliff and start looking for something more….titilating. The SV echo-chamber ensures your options are all the same.

(You saw it coming) We’re different. You might write off early stage companies because they contain outsized risk for the same job, but life is only as good as the unique experiences you chase. Novelty is risky, but so is a fun life. And you’re wrong that all early companies have outsized risk for the same job - this role is not your typical CV Researcher/Engineer. At Struction, you will be THE chief CV witch/wizard leading a novel architecture for harnessing latent intelligence to perform inference on the most semantically-rich imagery possible: construction drawings. Oh, and you get to see your work literally transform the world - the REAL world.

There is a chance we fail, but you’ll still get to build something way cooler than your friends working on agentic wordcels. When we succeed, your work will be renown for breaking the agility barrier of reality into a level only known previously by the digital realm.

Build something IRL using your digital wizardry

Struction’s mission is to replace today’s compute modality on the build-side of construction: blueprints (architectural / construction drawings). Architects and engineers already work in digital (though semantically poor) environments. Regardless, their output - 2D drawings and narratives - still serve as the technical language on the jobsite.

The worst part about using 2D drawings (PDF’s) is the time it takes to familiarize and integrate with a new project. That’s why Struction’s first move is to automate the onboarding process for every subcontractor and supplier and replace it with something dimensionally richer: 4D simulation. No one is doing this because (1) they think BIM has reached saturation (it hasn’t), (2) they think contractors don’t buy software (they do), or (3) they think it’s too hard (until us).

If we succeed, we will be the proprietors of THE compute platform in construction, replacing the job of drawings on jobsites today. Struction will lead the renaissance of iterating in reality. You get to write code that changes the REAL world - not the Nᵗʰ copilot for faster dev cycles or the AI girlfriend.

We’re based out of Colorado - one of the last places where reality, and productivity, still matter (+1 for quality of life). We’re close to our customers and intent on staying close as a company. We believe in working IRL but are consistently flexible. Getting your s*** done is more important than hours in the office, but time with your teammates is priceless. We lift, we bio-hack, and we love our work.

This summer we raised a pre-seed of over $800k from Eniac Ventures, Soma Capital, and DayDream Ventures. We’ve launched, we have customers / users, and we’re growing quickly. We’ll be working out of the Denver Metro area.

Responsibilities

Struction’s lead computer vision researcher / engineer will transition to lead the architecture of Struction’s ML vision models, model training, and computer vision inference.

1. Architect multi-modal, multi-step tasking of trained, fine-tuned, and adopted ML models on a previously untouched dataset of semantically rich imagery.
2. Ideate, optimize, and implement new architectures to capture latent intelligence of cutting-edge models and improve reliability, performance and efficiency.
3. Design machine learning system architecture that can abstract and automate critical machine learning product lifecycle including training, model management, tracking, cost management, and deployment.

Minimum Qualifications

1. You have developed cutting-edge, novel models for computer vision or multi-modal tasks like semantic segmentation, object detection, instance segmentation, text-to-image, etc.
2. You have worked with large amounts of real-world data, built data pipelines, and designed model architectures to productionize models. You have rapidly executed complex experiments to evaluate models in production environments.
3. You understand machine learning methods, linear algebra, and statistics; you have been through the full ML model lifecycle and solved challenges at each stage in development.