AI Engineer — Design Intelligence (Remote — Worldwide)

As an AI Engineer in the Design Intelligence group, you will build the foundation for generative design experiences that augment human creativity rather than replacing it. You will tackle complex problems at the intersection of computer vision, large language models, and vector graphics to automate tedious workflows and spark inspiration. Your work will directly impact millions of designers by shipping production-grade AI features like FigJam AI, intelligent layer organization, and semantic search across the Figma spatial canvas.

- Develop and deploy generative AI models that understand and generate Figma's internal scene graph and design tokens. - Design high-performance infrastructure to handle real-time inference on the web using technologies like WebAssembly or specialized edge workers. - Collaborate with product designers to create intuitive 'human-in-the-loop' interfaces for AI-assisted design. - Optimize RAG pipelines to enable context-aware suggestions based on a team’s unique design system and component library. - Evaluate and benchmark model performance specifically for design-oriented tasks, ensuring high visual fidelity and brand consistency. - Partner with the security and legal teams to implement robust data privacy standards for training and fine-tuning. - Contribute to Figma’s shared AI infrastructure, enabling other product teams to integrate intelligent features seamlessly.

- 5+ years of software engineering experience with a focus on machine learning or data-intensive applications. - Strong proficiency in TypeScript/React and experience integrating ML models into production web environments. - Practical experience fine-tuning and prompting Large Language Models (LLMs) and Multi-modal models (LMMs). - Deep understanding of vector databases (e.g., Pinecone, Milvus) and RAG (Retrieval-Augmented Generation) architectures. - Experience with Python-based ML frameworks such as PyTorch, JAX, or LangChain. - Background in computational geometry or manipulating complex tree-based data structures like SVG or document object models. - Proven track record of shipping user-facing AI features at scale in a collaborative, remote-first environment.

Equity, healthcare, parental leave.