Romain Huet is OpenAI’s head of developer experience, shaping how developers build with its models, APIs and Codex. The French entrepreneur previously co-founded Jolicloud and led developer-platform initiatives at Twitter and Stripe, building a career around making complex infrastructure useful to other engineers.
Huet and Tariq Krim founded Jolicloud in 2009, developing a cloud-based operating system and the Jolibook computer around accessible online services. He subsequently worked on Twitter’s developer platform before spending five years at Stripe, leading developer-platform product management and global developer relations. His 2018 introduction of Stripe’s developer dashboard consolidated API monitoring, error investigation, webhook diagnostics and integration upgrades.
After joining OpenAI to lead developer experience, Huet applied that platform-building experience to multimodal AI and increasingly autonomous coding tools.
- Multimodal interfaces that preserve context. Huet emphasizes what unified text, vision and audio make possible: lower-latency conversations, natural interruptions and less context loss than voice systems assembled from separate models. His GPT-4o demonstrations included diagnosing responsive-layout problems in a React and Tailwind interface through shared-screen conversation.
- Developer infrastructure matched to real workloads. Huet argues that useful AI platforms need more than powerful flagship models: differentiated cost and capability, asynchronous batch processing, domain customization and dependable interfaces for connecting agents with tools and data.
- Open agent infrastructure with shared primitives. He advocates giving developers the same underlying components OpenAI uses itself: the Responses API, context compaction for long-running tasks, an open Codex harness and app server, extensible integrations, and interoperable AGENTS.md instructions. His explanation of Codex architecture also describes a harness capable of accommodating other models.
- Engineering beyond writing code. Huet frames autonomous coding systems as expanding engineers’ responsibility for problem definition, judgment, experimentation and product decisions. He has publicly highlighted both developers building with Codex and OpenAI’s international developer events, keeping developer adoption and feedback central to his work.
