SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Build is creating the agentic AI stack for the built world, helping institutional real estate teams automate complex development and acquisitions workflows. The company works with major alternative asset investors, developers, infrastructure owners, and public-sector partners to transform slow, document-heavy processes into AI-powered systems.
You will be a hands-on AI engineer responsible for building production agentic workflows that customers use in real estate development, acquisitions, diligence, planning, and project execution. This role sits at the intersection of product engineering and applied AI—not research, not demos, but shipped production software with measurable customer outcomes.
Key responsibilities include:
- Designing and shipping production AI workflows that reason across leases, zoning documents, site plans, drawings, financial models, market comps, investment memos, permits, and project history
- Working directly with customers and domain experts to map complex workflows into software systems with clear inputs, outputs, edge cases, and success criteria
- Owning full-stack product features end-to-end, from backend workflow logic to user-facing review, approval, and collaboration surfaces
- Building context strategies that help agents use the right documents, prior decisions, domain constraints, and tool outputs at the right time
- Creating robust retrieval, extraction, structured-output, and tool-calling flows for expert users
- Designing comprehensive evals for document understanding, grounded reasoning, workflow completion, visual QA, accuracy, latency, and customer usefulness
- Inspecting traces, debugging failures, improving prompts and workflows, and turning customer feedback into measurable system improvements
- Partnering with the core AI and infrastructure team to improve agent reliability, observability, and developer velocity
Example projects include building acquisition diligence workflows that produce grounded risk summaries with citations, visual reasoning workflows that identify feasibility issues, long-running development agents that track missing documents and route work to experts, and customer-facing review surfaces where experts can inspect reasoning and approve outputs.
Success in the first few months means shipping at least one production agent workflow used by customers or internal experts, with clear task success metrics, eval coverage, traceability, documented failure modes, cost and latency targets, and feedback loops for continuous improvement.
You should be a strong product-minded software engineer who has shipped production systems, built with LLM APIs, agent frameworks, structured outputs, tool calling, RAG, or document processing. You are comfortable with Python and modern backend systems, can move across the stack when needed, and care deeply about agent quality beyond prompts—including evals, traces, regressions, edge cases, latency, cost, and user trust.
