Handoff is the AI agent that runs a construction company. We help remodelers automate estimating, streamline operations, and win more work — backed by real-time cost data, intuitive design, and workflows that "speak contractor." With over 10,000 monthly active users and $6B in annualized project volume already flowing through our platform, we're becoming the trusted partner for the people who build our homes.

We are backed by $25M+ raised from Y Combinator, Initialized, and Greycroft. Our team is distributed across hubs in Austin, São Paulo, and Buenos Aires, and we are deeply focused on building intuitive, high-impact solutions that make a real difference for our users.

What you'll do

• Own our core AI metrics and the dashboards and alerts behind them.
• Run experiments on prompts, tools, context and models, from offline evals to controlled rollouts, and decide with data.
• Build the evals and observability that let the whole team ship AI changes safely.
• Improve what customers judge us on: accurate outputs, reliable actions and fast answers.
• Shape the architecture of our agents as the product grows: how context, tools and memory scale.
• Evaluate new models and providers as they are released, and lead migrations when they are worth it.
• Work across Handoff’s products and teams, wherever AI quality, speed or cost needs attention.
• Turn customer-reported AI problems into diagnosed causes and measurable fixes.
• Share what you learn and raise the bar for AI engineering across the team.

About You

• You’ve owned an LLM-powered feature or agent in production and can point to a metric you moved.
• You design experiments carefully and can tell a real improvement from noise.
• 5+ years of software engineering experience, with strong Python and working knowledge of TypeScript/Node, SQL and cloud-based architectures.
• Practical experience with agents: tool calling, context management, retrieval and structured outputs, ideally with a framework such as LangGraph.
• Experience building evals and using tracing tools such as LangSmith, Langfuse or similar.
• You’re comfortable with ambiguity and switching between problems, and you set your own agenda from the data.
• You care whether the output is right for the customer, not only whether the model call succeeded.
• Proven experience working in a startup environment is a must.

What we offer

• 💸 Competitive salary in USD
• 💰 Attractive stock options
• 🌴 Unlimited PTO
• 🚛 Relocation allowance
• 👨‍💻 Top-notch equipment
• 🧳 Team offsites around the world!