Jimmy Lai announces he is in charge of OpenAI ChatGPT/Codex Web and Desktop infrastructure team
On November 9, Jimmy Lai posted on his personal X account that he is now in charge of OpenAI’s ChatGPT/Codex Web + Desktop infrastructure team. He said that if anyone had previously significantly accelerated other developers in the areas of platform, tools, reliability, or developer experience (DX), there is now an opportunity to do similar work at AGI scale, and he welcomed interested people to contact him.
In later replies to the same post, he added that part of the work in this role includes helping to implement features that Andrew Ambrosino wants to build. Jimmy Lai previously served as Director of Engineering for Next.js and React at Vercel, and also worked on React Native at Meta; in September 2026, he posted that he had joined OpenAI as an individual contributor to work on Codex. His current personal profile describes him as an engineering manager for OpenAI’s codex/chatgpt web infra.
This information comes from public posts made by Jimmy Lai himself, and OpenAI has not separately issued an official announcement for the corresponding team or recruitment.
Original source:
Title: btw, am now leading ChatGPT/Codex Web + Desktop infra group... Publisher: Jimmy Lai (@feedthejim) Publication date: 2026-11-09 URL: https://x.com/feedthejim/status/2119832776434393225 Source type: firsthand post Verification status: Read the full original post and subsequent replies, personal profile, and related historical posts on 2026-11-09 Supporting content: Supports his claim that he is now in charge of the ChatGPT/Codex Web + Desktop infrastructure team, that he is recruiting people with experience in platform/tools/reliability/DX to work at AGI scale, and the statement about helping implement @ajambrosino’s features; does not prove an official appointment or the specific size and scope of the team’s responsibilities
ABAB AI Insight
Before joining OpenAI, Jimmy Lai had long worked in front-end and developer tools. At Vercel, he served as Director of Engineering for the Next.js and React team, driving large-scale framework and toolchain iterations; earlier, he worked on React Native at Meta. In September 2026, he moved to OpenAI as an individual contributor working on Codex. In this post, he describes his role as the head of the ChatGPT/Codex Web and Desktop infrastructure team, reflecting a shift from individual contribution to infrastructure coordination.
OpenAI has continued in recent months to advance the integration of ChatGPT and Codex across desktop and web. Andrew Ambrosino is responsible for development related to the Codex desktop app, and in public interviews has described building Codex into a hub for knowledge work while emphasizing near-universal weekly internal use. In his replies, Jimmy Lai directly linked his role to helping deliver features proposed by Ambrosino, indicating that the scope of the infrastructure team has extended to supporting rapid product experimentation and delivery, rather than only maintaining underlying stability.
Compared with Anthropic’s path for Claude Code, where Boris Cherny moved it from an internal prototype to a public product, OpenAI’s Codex has likewise expanded from an early internal tool into a core component of the desktop app. The similarity is that both focus on improving developers’ real iteration speed; the difference is that Jimmy Lai explicitly announced recruitment under the banner of an “infrastructure team,” directly tying it to the product lead’s feature needs. This arrangement transfers the developer-experience optimization experience accumulated at Vercel and Meta into the web and desktop runtime environment at AI-agent scale.
This role setup changes OpenAI’s engineering resource allocation for the ChatGPT and Codex desktop sides: having someone with large-scale tool and platform experience lead infrastructure can directly support product-side requirements for long-running tasks, cross-device synchronization, and reliability. A measurable indicator will be the extent of subsequent public-product improvements in long-running task stability and the switching experience between Web and Desktop.
ABAB News · Cognitive Law
When an infrastructure lead recruits publicly, the role is directly tied to the product lead’s feature needs.
Experience in front-end frameworks and toolchains is being transferred to AI-agent web and desktop infrastructure.
A personal post announcing a role and recruiting does not mean an official team announcement has been issued.