Software Engineer, ChatGPT Infrastructure

OpenAI•San Francisco, CA

16h

About The Position

ChatGPT is a rapidly evolving system: new capabilities ship continuously, product surfaces change quickly, and usage patterns shift week-to-week. Supporting that pace requires infrastructure that can handle real production constraints—high concurrency, unpredictable traffic patterns, complex dependency graphs, and frequent change. The ChatGPT Infrastructure team builds and operates the platforms that enable fast iteration without compromising performance or reliability. We design shared systems, data paths, rollout mechanisms, and reliability guardrails that teams rely on to ship changes to ChatGPT at scale. We focus on high-leverage infrastructure: primitives and “golden paths” that incorporate operational lessons as defaults, so engineers don’t need to rediscover failure modes, latency pitfalls, or integration issues each time they build something new. We’re hiring Senior and Staff Engineers to design and build infrastructure systems that underlie ChatGPT and multiply the effectiveness of teams building user experiences. This is not a support-only role. It’s a platform-building role: you’ll define interfaces, develop core abstractions, and create tooling to make safe, fast iteration the norm. Your work will reduce friction, prevent regressions, improve performance, and ensure systems scale gracefully as the product grows.

Requirements

Experience building and operating large-scale distributed systems in production (high throughput, concurrency, and failure handling).
Strong fundamentals in systems design, including caching, consistency, queueing/backpressure, and resilient dependency management.
Ability to reason about performance (latency distributions, tail behavior, bottlenecks) and translate analysis into concrete engineering work.
Track record of building platforms or shared infrastructure that improves velocity and correctness for other teams.
Excellent communication and collaboration skills—aligning on interfaces, navigating tradeoffs, and driving cross-team execution.

Nice To Haves

Experience designing paved roads / golden paths (frameworks, libraries, self-serve tooling) that shape engineering behavior at scale.
Deep understanding of reliability techniques: graceful degradation, circuit breakers, load shedding, rate limiting, and fault isolation.
Experience building systems for safe iteration: progressive delivery, correctness checks, automated rollout gates, and production validation.
Strong instincts for API and contract design—how to create interfaces that are stable, evolvable, and hard to misuse.
Prior work that demonstrates “force multiplier” impact: enabling many teams via a small set of well-crafted primitives.

Responsibilities

Work on Platform foundations & frameworks: Core libraries, service frameworks, and shared components that standardize system building, integration, and evolution.
Work on Scalability & performance primitives: Patterns and infrastructure that reduce tail latency, improve throughput, and keep costs predictable as demand increases.
Work on Reliability guardrails: Mechanisms that prevent outages by design—rate limiting, load shedding, dependency isolation, backpressure, safe fallbacks, and robust regression controls.
Work on Developer productivity via golden paths: Paved roads for common workflows (data access patterns, service integration, request lifecycle) that are fast, safe, and easy to use.
Work on Observability & debugging systems: Instrumentation, metrics models, and investigative tooling that turn vague symptoms (“it’s slow”) into precise, actionable diagnoses.
Work on Safe change management: Deployment and rollout systems that support rapid iteration with confidence—progressive delivery, automated verification, and fast rollback strategies.
Work on Interface and contract design across boundaries: Clean APIs and stable contracts that reduce coupling and enable independent evolution across a complex ecosystem.
Build and evolve infrastructure platforms used by many engineers and services.
Translate real-world constraints into clean abstractions: simple APIs, enforceable contracts, safe defaults.
Drive improvements in reliability and performance through principled design, measurement, and iterative hardening.
Partner with engineering and product teams to identify systemic pain points and develop reusable solutions.
Own outcomes end-to-end: design → implementation → rollout → operational maturity.