Founding Engineer - Arkhe

Over time, these systems will evolve into a persistent intelligence layer that can help teams understand, design, coordinate, review, and ultimately build the physical world.

You will:

• Develop the core intelligence behind autonomous agents for reviewing, designing, and coordinating the built world.
• Create physical-AI systems that understand BIM models, 3D geometry, spatial relationships, and real-world constraints.
• Build the persistent project intelligence layer that allows AutoSitu to understand one project continuously as it moves from feasibility through design and construction.
• Design distributed workflows that coordinate specialized agents across multimodal project artifacts in parallel.
• Establish the context and data infrastructure for city-scale digital twins that autonomous agents can understand, design, and act within.
• Create evaluation infrastructure for high-stakes outputs where accuracy, evidence, and judgment all matter
• Improve agent reliability through structured reasoning, retries, observability, human feedback, and failure analysis
• Own product features end to end—from system architecture and model behavior to UI, deployment, and customer feedback
• Work directly with architects, engineers, developers, contractors, and public-sector reviewers to understand how the product performs on real projects
• Make foundational decisions about our engineering architecture, product direction, and technical culture

What to expect in your first 90 days

0–30 days: Learn the product, customers, review workflows, and core AI architecture. Begin shipping improvements to the agent and evaluation infrastructure.

30–60 days: Own a meaningful product or intelligence workflow end to end, from technical design through deployment and evaluation on active projects.

60–90 days: Lead the development and customer rollout of a core capability used in real development-review workflows.

You will not spend your first several months waiting for a roadmap or being assigned isolated tickets. You will be expected to understand the problem, make decisions, build the solution, and learn directly from how customers use it.

We are looking for someone who

• Has built and shipped technically ambitious products
• Wants to work across AI systems, backend infrastructure, product, and frontend when necessary
• Understands that reliability, traceability, and evaluation matter more than impressive demos
• Thinks carefully about concurrency, state, retries, idempotency, rate limits, and failure modes
• Is deeply curious about frontier multimodal and agentic architectures
• Can turn ambiguous, real-world workflows into clear technical systems
• Wants to speak directly with customers and understand the domain rather than build from a distance
• Moves quickly while maintaining strong engineering judgment
• Is comfortable entering a complex industry and learning from experienced practitioners
• Wants substantial ownership and responsibility at an early-stage company
• Cares about improving how housing, infrastructure, and communities get built

Experience in architecture, construction, permitting, or government is not required. A willingness to learn the domain deeply is.

Our technology

We are building distributed, multimodal AI systems that orchestrate specialized agents across technical documents, BIM models, spatial data, and 3D/4D environments.

Our platform combines multimodal foundation models, agentic vision, multi-agent workflows, BIM-native intelligence, spatial reasoning, a knowledge-graph context engine, document retrieval, evidence localization, and rigorous evaluation infrastructure.

Our stack includes TypeScript, Next.js, Mastra, Postgres, Supabase, Inngest, cloud infrastructure, and modern language and vision models.

We care less about whether you have used every tool in our current stack and more about whether you can reason from first principles, choose pragmatic technologies, and build systems that work reliably for real users.

How we work

Ownership over narrow roles. Early team members own problems end to end rather than operating within rigid functional boundaries.

Stay close to the customer. Everyone learns directly from the people using the product. Customer conversations are part of product development, not a separate function.

Evidence over intuition. We test our systems on real projects, build evaluations, investigate failures, and improve based on what the evidence shows.

Speed with judgment. We move quickly, but we do not confuse unnecessary complexity with technical sophistication.

Low ego, direct communication. We challenge ideas openly, acknowledge mistakes quickly, and prioritize the quality of the outcome over who proposed it.

Mission-oriented. We are building infrastructure for how the physical world gets reviewed and approved. The work is difficult, but the potential economic impact is enormous.

Why join

You will join early enough to shape the product, architecture, culture, and direction of the company.

You will work directly with the founders, ship into active projects, interact with customers making real development decisions, and solve applied-AI problems that do not have established playbooks.

The opportunity is larger than automating individual plan checks. We are building the intelligence layer through which cities and development teams understand, design, review, coordinate, and ultimately build the physical world.

A meeting lands on your calendar

If they’re a yes, I book the chat. You show up — that’s the whole job-hunt.