SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
PermitFlow is an applied AI company building AI agent infrastructure for construction permitting and compliance workflows. The company serves the nation's builders, helping them navigate permitting, licensing, inspections, and project closeouts. PermitFlow has raised over $90M (Series B) from top-tier investors including Accel, Kleiner Perkins, Y Combinator, and others, and has already delivered impact across $20B+ in development.
You will own the design and implementation of PermitFlow's core product AI Assistant, agentic infrastructure, and workflow agents. This role requires both UX sensibility and deep technical acumen to deliver high-quality agents in a complex domain (construction permitting) while also building the foundational agentic infrastructure and data flywheel that scales across the organization.
Key responsibilities:
- Own the agent architecture: design the reasoning loop, planning and decomposition strategy, state and session model, and recovery behavior for long-running tasks.
- Build the context layer: manage retrieval, memory, and context-window optimization over project documents, jurisdictional requirements, prior submissions, and customer-specific conventions.
- Design the action surface: define tools the assistant can call, permissions models, and human-in-the-loop checkpoints for expensive or irreversible actions.
- Make evaluation the engine of progress: build offline and online eval harnesses that measure task completion and failure modes; establish the iteration loop for continuous agent quality improvement.
- Set the bar for the org: establish abstractions, SDKs, and code review standards that other engineers use to build agentic features.
- Stay close to customers: observe permit coordinators and builders using the assistant in real conditions, and incorporate learnings into system design.
The role reports directly to the VP of Engineering. PermitFlow's HQ is in New York City with a hybrid schedule (3 in-office days per week).
REQUIREMENTS:
- 10+ years of software engineering experience with recent time spent building LLM-powered products.
- Shipped an assistant or agentic product at meaningful scale.
- Deep, current knowledge of the agentic stack: tool calling and tool design, context engineering, memory, retrieval, orchestration patterns, etc.
- Built eval harnesses that changed what a team worked on; can articulate what makes an eval trustworthy.
- Genuinely strong software engineer (not just an AI specialist): owned services in production, writes code others are glad to inherit.
- Operate with earned authority: influence through demonstrated judgment, working code, and clear writing rather than title or mandate.
About PermitFlow
Real Estate / Construction Tech; SaaS / Enterprise Software — construction permitting workflow software.
