engblogs

A technical breakdown of OpenAI's response to the TanStack npm supply chain attack, detailing certificate rotation, notarization revocation, and coordinated macOS app updates to safeguard users while preserving provenance and integrity of software supply chains.

An in-depth look at building an elevated Windows sandbox for Codex that balances effectiveness and safety using synthetic SIDs, write-restricted tokens, and firewall-based network controls across a multi-binary architecture.

Viaduct 1.0 marks Airbnb’s shift from an internal tool to a community-driven, production-ready data mesh built on a GraphQL-based, multi-tenant architecture with a central schema and governance, designed to complement GraphQL Federation at scale.

Viaduct 1.0 transitions Airbnb’s internal data tooling into a community-driven, production-ready GraphQL-based data mesh that enables decentralized, multi-tenant data APIs with a central schema governance model.

Modular: Translating to Mojo via AI Agents shows how AI agents translate CUDA/Triton GPU kernels to Mojo, delivering portable, high-performance Mojo code and a practical path to Mojo 1.0 with open-source tooling and agentic workflows.

AutoScout24 scales engineering by embedding Codex and ChatGPT into AI-powered workflows across engineering, data, and product teams to accelerate delivery, improve code quality, and expand AI adoption at scale.

NVIDIA engineers and researchers accelerate end-to-end machine learning workflows by using Codex with GPT-5.5 to turn research ideas into autonomous, production-ready systems.

Open, agent-driven ML challenges like Parameter Golf show how AI-assisted research accelerates experimentation, evaluation, and talent discovery under tight constraints.

Discover how finance teams transform inputs from dashboards, forecasts, and owner notes into review-ready assets with ChatGPT Work, no coding required and focused on telling the right story.

AutoScout24 scales engineering with AI-powered workflows by embedding Codex and ChatGPT across 2,000 employees to accelerate development, improve code quality, and expand AI adoption.

NVIDIA engineers and researchers use Codex with GPT-5.5 to accelerate end-to-end machine learning workflows, ship production systems, and surface insights across research, development, and deployment.

AutoScout24 scales engineering with AI-powered workflows by pairing enterprise-wide ChatGPT access with Codex-powered coding agents to accelerate development, boost code quality, and broaden AI adoption across 2,000 employees.