Design and build LLM-powered AI agents that automate support workflows across Zendesk, Intercom, and internal systems.
Develop agents that understand requests, plan next steps, retrieve relevant context, take action, and elevate to human agents when needed.
Own prompt engineering, tool-use design, and evaluation frameworks to ensure reliable, production-quality agent behavior.
Build and maintain the MLOps infrastructure behind it (model serving, CI/CD, monitoring, safe rollouts).
Track impact through clear KPIs such as resolution time, first-touch resolution, and automation rate.
Integrate these systems with our existing TypeScript/Python stack, internal APIs, and knowledge bases.
Work closely with Support, Product, and Engineering to identify the highest-impact automation opportunities.
Share what you learn with the team and help other engineers apply LLM-driven automation effectively.
Your profile
6+ years of experience in Software or Machine Learning Engineering.
Proven track record building and deploying LLM, GenAI, or agentic AI systems in production.
Experience designing AI systems that solve real business problems at scale — ideally in workflow automation, decision support, or autonomous task execution.
Hands-on expertise with LLM application architecture: tool calling, RAG, orchestration, memory/context management, evaluation, and guardrails.
Solid background in MLOps and production AI infrastructure (deployment, CI/CD, monitoring, evaluation, incident handling).
Proficiency in Python and/or TypeScript/JavaScript.
Excellent English (C1); German is a plus.
Why us?
High impact and ownership from day one — your work will directly shape deeploi's AI strategy.
A small, ambitious team in Berlin and Munich — backed by top European VCs including Atomico and Cherry Ventures.