AI Engineer

LemonLime is automating sales and marketing for small businesses. We're looking for an AI Engineer to help build the systems that make that possible.

A small business’s goal from LemonLime starts as vague as “get me customers.” Turning that into actual results for them requires much more than a single model call or workflow. The system needs to understand the team, the business, the industry, the competition, the trends, the meta, all of it, and from there: determine what strategies make sense, gather external information, generate outputs, evaluate its own work and run simulations, and continue to learn over time.

This means getting a system to do things correctly, autonomously, and repeatedly for thousands of different businesses.

What you'll do

• Agent orchestration, long-running workflows, retrieval and context management, structured generation, tool use, evaluations, data pipelines, and production infrastructure.
• Combine model reasoning with deterministic software (seriously, this is a huge component)
• Integrate models with APIs, external data sources, browsers, databases, and third-party tools.
• Ship constantly, adapt to flows that actually work for real customers (user feedback through observed behavior is law), and rebuild and design what isn’t working.
• Work across the stack when necessary. There is no boundary where “the AI work” ends and traditional software engineering begins, we simply do too much across the board too fast to define those lines.

You might be a fit if

• You have incredibly strong fundamental understanding of software development and architecture outside of using AI
• You know how to and have experience with using AI to build (not just Claude Code – the step above that)
• You're up to date with models, tools, agents, software, the benefits, tradeoffs, risks for each, etc
• You move extremely fast and can ship at speed with minimal failure risk

Our technology

We combine frontier models with deterministic software; sometimes, we just have to bake things in ourselves.

Our work touches agent orchestration, retrieval, structured LLM outputs, evaluation systems, long-running jobs, data pipelines, third-party integrations, and large-scale model inference. Our broader stack includes modern TypeScript/Python services, Postgres/Supabase, Vercel, PostHog, and a growing ecosystem of APIs and external data sources.

We deliberately know not everything should be or needs to be an agent.