Founding AI Engineer

Mason AI; San Francisco (CA); On-site; USD 160,000 - 210,000; Full time

Benefits offered by this job

Competitive salary

Equity (early-stage)

Free lunch & dinner

Health insurance (HSA)

401(k)

Commuter benefits

Wellness benefit

Impact on country

Job summary

Mason AI is seeking a Founding AI Engineer to improve accuracy on evals and push the frontier of multimodal AI agents that verify blueprints for building code compliance. You’ll own experiments, autoresearch, and feature sweeps to shape our product from day one.

You’ll work deeply in AI for legal reasoning, spatial understanding, and visual understanding, with in-person collaboration in California. This role directly impacts the built environment and the country’s construction capabilities.

Qualifications

• SWE with a drive to grow in ML/AI engineering
• Proven ability to run experiments quickly and prototype
• Ownership mindset and comfort with ambiguity

Responsibilities

• Improve accuracy of multimodal agents that check blueprints for building code compliance
• Run evals, autoresearch, and feature sweeps to advance models
• Collaborate with users and iterate on product features

Job description

Mason AI is on the frontier of agentic AI for the built world: we're the technical leaders at building AI that checks blueprints are buildable and safe.

We're looking for a Founding AI Engineer to improve Mason's accuracy on our evals. You will go deep on the cutting edge of optimizing multimodal AI agents. We have similar technical problems as other agentic AI startups, stacked on top of AI for legal reasoning, spatial reasoning, and visual understanding. Your work will directly affect the built environment of California within the first month of joining.

You will have a massive impact on the product, the company's future, and the country's, as we accelerate American construction. Join us.

Who you are

• Either a SWE passionate about growing as an ML/AI Engineer, or a ML Engineer who's fast at prototyping.

• You can run experiments quickly and develop research taste.

• You're mission-oriented and looking for your life's work. You care about America's ability to build.

• You love ownership and don't mind ambiguity or unblocking yourself.

• You have the curiosity of a policy wonk to digest building codes and architectural diagrams.

What you'll do

• 60+% of your time will be spent improving the accuracy of our multimodal agents that check blueprints for building code compliance. This can range from building autoresearch to improving our evals to running sweeps on new experimental features you build. Your research taste will direct this work.

• Talk to users and dogfood the product weekly

• Wear hats as needed on topics ranging from running computer vision experiments, lightweight data engineering, building internal tooling, improving the webapp, and more.

Our operating principles

• Fail fast and cheap. Slugging percentage matters more than batting average.

• We can make a lot of mistakes as long as we get a few things right.

• A written culture - we document our thinking on the few things that matter most.

• Internal tooling is a first-class use of time.

• In-person work is a must.

Benefits

• Competitive salary ($160k-$210k).

• Generous early-stage equity (0.75%-2%).

• Free lunch and dinner (ordered at 6 pm) at the office.

• Fully paid health insurance with over $1000/year in employer contributions to an HSA.

• A 401(k) that accepts pre-tax and Roth contributions.

• A commuter benefits account and a monthly wellness benefit.

• Opportunity to shape the future of our country.

Interview process

Our interview process aims to be lightweight and respect candidates' time. We expect it to take 2 weeks or less.

• Intro call (30 min) - Informal chat on your interests and experience

• AI Coding & Culture interviews (90 min)

  • Designed to mirror real work you'd do at Mason, you'll build a prototype for a slice of our product. AI coding tools (e.g. claude code, conductor, wisprflow) are strongly encouraged.

  • Learn more about how you've navigated challenges in (or outside) of work.

• Paid work trial (3 days) - We'll set you up to run evals for our multimodal agent on our repo, and we'll ask you to generate hypotheses for experiments to run, run those experiments, and then analyze them. You'll be evaluated on your speed at running thoughtful experiments. We're flexible on scheduling, and usually run these over the weekend to minimize impact on your weekday schedule.

• Full-time offer