Forward Deployed Engineer – Agentic Platform (Seoul)

About the role

Cohere is building a secure AI workspace platform called North that lets enterprises deploy large language model (LLM) agents while keeping data under control. As a Forward Deployed Engineer you will work directly with customers to design, build, and launch agentic workflows that solve high‑impact business problems across sectors such as finance, healthcare, and telecommunications.

Key responsibilities

• Collaborate with enterprise clients to translate ambiguous business challenges into well‑defined agentic problems and success criteria.
• Lead the end‑to‑end design, implementation, and production launch of LLM‑powered agents that reason, plan, and act across tools and data sources.
• Iterate rapidly while ensuring enterprise‑grade reliability, observability, safety, and auditability of deployed agents.
• Travel up to 20‑40% to work on‑site with customers and partners, providing post‑deployment support and continuous improvement.

Required profile

• Proven experience building and operating production‑grade software systems, especially LLM‑based agents in real‑world settings.
• Strong technical leadership, having guided architecture and delivery of complex AI solutions for customers.
• Excellent written and verbal communication in both English and Korean.

Required skills

• Python and/or JavaScript/TypeScript.
• Deep familiarity with modern LLM ecosystems (e.g., GPT, Claude, Gemini), vector databases, Retrieval‑Augmented Generation (RAG) architectures, and agent/orchestration frameworks.
• Prompt engineering, multi‑step agent workflows (e.g., ReAct), tool integration, and robust failure handling.

What we offer

• Weekly lunch stipend of $75 (or equivalent local currency).
• Full health, dental, and mental‑health benefits plus education & learning stipend.
• Retirement plans (RRSP matching, 401k, pension), 100% parental‑leave top‑up for up to 6 months.
• 6 weeks paid vacation, travel budget for office visits, and $500 home‑office setup stipend.