Applied AI Engineer — Agentic Systems at Rozeta Labs
Job Description
You are a strong engineer who is excited by applied AI, not just model benchmarks.
You can build useful software quickly.
You understand LLMs, agents, retrieval, tool use, structured outputs, evals, and orchestration — but you also understand that the hard part is usually the business process, not the model call.
You are comfortable working in ambiguous environments where the requirements are messy at first.
You can talk to non-technical operators, translate what they mean, and build what they actually need.
You care about shipping systems that survive contact with the real world.
Useful Experience
- Strong experience with TypeScript, Python, or both.
- Experience building with LLM APIs such as OpenAI, Anthropic, Gemini, or similar.
- Experience with AI agent frameworks, workflow orchestration, structured outputs, retrieval, vector databases, tool calling, and evals.
- Experience integrating with third-party APIs, webhooks, databases, CRMs, internal tools, and SaaS platforms.
- Experience building internal tools, dashboards, admin panels, approval flows, or human review queues.
- Experience with cloud platforms such as Cloudflare, AWS, GCP, Azure, Supabase, or Vercel.
You’ll Be a Great Fit If
- You like turning vague business pain into working systems.
- You enjoy talking to operators and figuring out what is actually broken.
- You are not precious about tools, frameworks, or hype cycles.
- You can move fast without being reckless.
- You are comfortable building the first version, watching it break, learning why, and making it better.
- You understand that production AI requires guardrails, evals, logging, observability, permissioning, and human escalation.
- You care more about business outcomes than clever architecture.
- You would rather ship something useful than debate whether it counts as “agentic.”
You May Not Be a Fit If
- You only want to work on model research.
- You mainly want to build demos.
- You think a chatbot is the answer to every workflow problem.
- You need perfectly written requirements before you can start.
- You are uncomfortable talking directly to clients or operators.
- You are allergic to messy systems, legacy tools, spreadsheets, weird APIs, or business logic that lives in someone’s head.
