Drimva

Figma

AI Engineer, Internal Tools

$153K–$296K

San Francisco, CA • New York, NY • United States +2 · Full-time · Posted Oct 27, 2026

Location
San Francisco, CA • New York, NY • United States
Canada
US
Experience
5+ years of software engineering experience.

About the job

Build and operate production AI agents, assistants, and workflow automations that address business needs across Figma, integrating them with enterprise systems and data.

What you'll do

  • Partner with business teams to find high-value AI opportunities and develop solutions from discovery through production.
  • Build evaluations, feedback loops, safeguards, and reliable integrations for AI workflows that may take business-critical actions.
  • Identify reusable patterns and develop shared capabilities with internal AI platform and infrastructure teams.

What they ask for

  • 5+ years of software engineering experience building production applications, integrations, backend systems, or AI-powered products.
  • Hands-on experience building applications with LLMs, agents, retrieval-augmented generation, function calling, schema-constrained outputs, or evaluation frameworks.
  • Experience taking an AI application or workflow beyond a prototype and operating it in production.
  • Strong API, integration, system design, and production reliability fundamentals; experience translating ambiguous business problems with non-engineering stakeholders.

Nice to have

  • Background in internal AI applications, Applied AI, Forward Deployed Engineering, or AI Solutions Engineering.
  • Experience with MCP, agent orchestration, AI platforms, enterprise systems, evaluations, identity and access patterns, or enterprise governance.

Pay

$153,000 — $296,000 USD (base salary)

Tools

LLMs, RAG, MCP, APIs, AWS Bedrock, AgentCore, OpenAI, Anthropic

AI focus

Develop and productionize internal AI agents, assistants, and workflow automations; evaluate them and design safe controls for enterprise actions.

Model-written summary · the original posting is authoritative

First seen Oct 28, 2026 · Last verified Nov 4, 2026 (UTC)