Software Engineer

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

Benefits offered by this job

Free lunch and dinner in the office

Fully paid health insurance

HSA contributions

401(k)

Commuter and wellness benefits

Job summary

Berkley Hunt is partnering with an early-stage startup building agentic AI for the built world. Their platform uses multimodal AI agents to check blueprints for building code compliance, helping to accelerate construction and tackle the housing crisis.

Working closely with the founding team, you'll focus on improving the accuracy of their multimodal agents, tackling challenges at the cutting edge of agentic AI, spatial reasoning and computer vision.

Qualifications

• 3+ years of software engineering experience with the ability to make mature architectural decisions on core systems.
• Hands-on experience using AI coding tools and parallel agent workflows.
• Mission-driven mindset with a genuine interest in improving housing and construction.
• Comfortable with ambiguity, high ownership, and the ability to unblock yourself.
• Pragmatic, experimental approach valuing speed and learning from failure.

Responsibilities

• Improve the accuracy of multimodal agents that check blueprints for building code compliance, including post-training models and evals.
• Build new agents to extend the product into commercial and multifamily housing across the US.
• Contribute across the stack from data engineering and tooling to improving the web application.
• Set up and optimize the team's AI coding agents and workflows.
• Engage with users weekly to understand needs and shape the product.
• Foster a strong written engineering culture as the company scales.

Skills

AI coding tools

Parallel agent workflows

Software architecture

Housing impact

Ownership

Ambiguity tolerance

Experimentation

Job description

Berkley Hunt is partnering with an early-stage startup building agentic AI for the built world. Their platform uses multimodal AI agents to check that building blueprints are buildable, safe and compliant with building codes, helping to accelerate construction and tackle the housing crisis.

Working closely with the founding team, you'll focus on improving the accuracy of their multimodal agents, tackling challenges at the cutting edge of agentic AI, spatial reasoning and computer vision. This is a true founding engineering opportunity, with your work directly shaping real-world construction projects within your first month.

What You'll Do

• Spend the majority of your time improving the accuracy of multimodal agents that check blueprints for building code compliance, including post-training models and running experiments against evals
• Build new agents that extend the product into commercial and multifamily housing across the US
• Contribute across the stack, from lightweight data engineering and internal tooling to improving the web application
• Set up and optimise the team's AI coding agents and workflows
• Speak with users weekly to understand their needs and shape the product
• Help establish a strong written engineering culture as the company scales

What We're Looking For

• 3+ years of software engineering experience, with the ability to make mature architectural decisions on core systems
• Strong hands-on experience using AI coding tools and parallel agent workflows
• A mission-driven mindset and genuine interest in improving housing and construction
• Comfortable with ambiguity, high ownership and unblocking yourself
• A pragmatic, experimental approach where speed and learning from failure are valued

Nice to Have

• Experience with multimodal AI, computer vision or spatial reasoning
• Experience post-training models or designing and running AI evaluations
• Background building internal tooling or developer productivity systems
• Onsite in California
• Salary: $160k-$210k plus generous early-stage equity (0.75%-2%)
• Fully paid health insurance, HSA contributions, 401(k), commuter and wellness benefits
• Free lunch and dinner in the office
• Lightweight interview process completed within around 2 weeks, including a hands-on AI coding interview and a paid 3-day work trial
• Founding engineering opportunity with significant ownership and impact