Blackbird Labs is a loyalty and payments platform establishing meaningful connections between the world’s best restaurants and their customers.

Served up in 2022 by Ben Leventhal, co-founder of Eater and Resy, Blackbird offers a fully customizable loyalty programming platform for restaurants and a consumer app designed to make every guest feel like a bonafide VIP no matter where they dine. Backed by Union Square Ventures, a16z, Amex Ventures, Variant Fund, and other forward-thinking investors, Blackbird is out to revolutionize the restaurant industry.

At Blackbird, our values guide everything we do and our talent principles define the key traits we seek in our employees.

We have passion for our product – committed to our mission and instinctively think like and go the extra mile for our customers. We are elite performers – striving for excellence, going above and beyond to achieve exceptional results, and learning quickly with horsepower, drive, and grit. We make others better – taking initiative to elevate others, giving and receiving feedback, and championing a culture of growth. We hold one another accountable to living out our talent principles as we march together to achieving our mission.

Think about how people dine: where they go, how often they come back, whose taste they trust, which rooms they have a relationship with. We're building Blackbird to answer those questions the way a well-connected friend would. We need a product-minded fullstack engineer to own it.

This is a hands-on IC role. Everyone builds. Everyone ships. The problems here don't stay in their lane. One fix might take you from a rendering bug to a prompt regression to a broken retrieval plan to a backend endpoint. You follow it wherever it goes.

Our team is not looking for the conventional AI resume or an exact stack match. We're looking for product judgment and first-principles thinking. Someone who cares about their craft.

• Build the chat experience diners talk to every day.
• Own the agent runtime end to end.
• Ship new tools the model can call, and the tests that pin their behavior.
• Make answer quality measurable: evals, tracing, and real user feedback.
• Improve retrieval so the agent recommends the right place.
• Take the product from answering questions to completing tasks.
• Keep turns fast, cheap, and reliable on mobile networks.
• Get work in front of real diners fast, with product and design.

• 5+ years shipping production software, owning features end to end.
• Able to debug across the stack: a streaming client, application code, SQL, and the boundaries between services.
• Real depth with LLM application development: tool calling, context management, retrieval, streaming.
• Strong React and TypeScript, fluency in Python, and willingness to work in Kotlin/Spring when the answer lives on the platform side.
• Solid Postgres and data modeling instincts, including a feel for what makes retrieval good or bad.
• Taste in evaluation: you know the difference between fixing a failing eval and gaming one.
• Product judgment about what makes an agent feel trustworthy, fast, and worth coming back to.
• Comfort with ambiguity and high ownership on a small team.
• You use current models seriously and have your own ideas about how agent systems should work.

• Agent evals and LLM observability tooling (Langfuse, Braintrust, LangSmith).
• Vector and geospatial search in Postgres (pgvector, PostGIS).
• Shipping AI into a native mobile surface: streaming, cold starts, bad networks.
• Frontier model APIs: tool use, prompt caching, extended thinking.
• Restaurants, hospitality, travel, or another domain where recommendations carry real-world consequences.
• Datadog, Amplitude, LaunchDarkly, AWS, or Terraform.