Prompt Engineer (Healthcare)

About us Talkie builds AI Agents for healthcare. Our agents handle patient–clinic communication end-to-end — voice calls, web chat, and SMS — so patients get 24/7 access to care and busy practices never miss a conversation. Every month, our AI Agents handle close to a million real patient conversations across the US and Poland. Trusted by primary care, specialty practices, and hospitals. The Role We're looking for a Prompt Engineer to own the quality and intelligence of our AI Agents — from prompt design to production. This is a high-impact role at the intersection of language, technology, and healthcare. Your work will directly shape how hundreds of thousands of patients experience care. This is not a research role. You'll be writing, testing, and iterating prompts that run live with real patients — so rigour, empathy, and a zero-error mindset matter as much as technical skill. What You'll Do - Design, write, and continuously optimise prompts that power our AI Agents — making them natural, accurate, and reliable. - Analyse real patient–agent conversations end-to-end, identifying failure patterns, edge cases, and opportunities to improve agent behaviour. - Propose and implement technical solutions around function calling, tool use, context caching, and other LLM capabilities that make our agents smarter. - Build and run evaluation frameworks to test agent performance before and after changes — because every conversation is with a real patient and there is zero margin for error. - Create clear, structured documentation and customisation instructions so that agents can be tailored to each client's specific workflows and needs. - Stay on top of the rapidly evolving LLM landscape — new models, techniques, and conventions — and bring the best ideas back to the team. - Work closely with the Product Manager (US market), engineering, and client teams to ensure agent quality across all deployments. What will you achieve with us? - Shape the experience that hundreds of thousands of patients have when they reach out to their doctor — across every channel — and make it better every single day. - Push the boundaries of what LLM-powered AI agents can do in a highly regulated, real-world, multi-channel environment. - Build evaluation and quality systems for conversational AI that don't exist yet — you'll be creating the playbook. - Have a direct, measurable impact on patient access to healthcare in both the US. Requirements This is a young and fast-moving field. We care less about years of experience and more about how you think, learn, and work. Must have - LLM Experience — Hands-on experience writing and iterating on prompts for production systems. More importantly, you learn fast — this field changes weekly and you keep up. - Analytical Rigour — Ability to review conversations, extract failure patterns, and turn findings into concrete, measurable improvements. - Communication & Collaboration — Comfort on client calls and working across technical and non-technical teams; you translate clearly in both directions. This is not a "sit in a cave and prompt all day" role ;). - Proactivity — You spot problems before being asked, flag them, and come with a proposed fix. - Zero-Error Mindset — Our agents talk to real patients. You understand the responsibility and bring the precision and care it demands. - English — C1+ English, strong written communication. Our agents talk to US patients. - Shifted Hours — You are available to work 12:00–20:00 CET at least 3 days per week (optimally 5) to overlap with US Eastern Time business hours. Nice to have - Function Calling & Tool Use - Experience with LLM tool-use patterns, structured outputs, and API integrations. - Evaluation Frameworks- Familiarity with eval tools — Braintrust, DeepEval, LangSmith, or custom pipelines. - Multi-Channel Experience - Understanding of voice AI nuances: latency, turn-taking, TTS/ASR — and how they differ from chat or SMS. - Healthcare Background - Prior work in healthcare, health-tech, or regulated industries where accuracy and compliance are non-negotiable. - Genuine Curiosity - You read release notes. You experiment with new models. You show up on Monday with fresh ideas. Your goals as a Prompt Engineer Short term — first 3 months - Develop a deep understanding of our AI Agent architecture, prompt patterns, client configurations, for our US market product. - Audit existing agent conversations, identify the top quality issues, and implement prompt improvements with measurable impact. - Take ownership of the agent testing and evaluation process — establish baselines and a repeatable QA workflow. - Get up to speed on our tooling (Langfuse, ClickUp, internal platforms) and the team's ways of working. Longer term — first 12 months - Own the end-to-end prompt and agent quality lifecycle across our US deployments. - Build and maintain a structured evaluation framework that catches regressions before they reach patie