Compiler Engineer - AI Inference

AI Systems Engineer - Codex Core Agents About The Team The Codex Core Agents team builds the agent harness that turns model capability into real-world action. We own the systems around the model: prompting and interpreting model outputs, executing actions safely in real environments, and feeding production experience back into better models and better agent behavior. This team sits close to research and works across the stack: harness, model interaction, inference, sandboxed execution, orchestration, evals, production reliability, and the performance envelope around tokens, latency, cost, capacity, and quality. The harness is open source and increasingly part of how models are trained and evaluated, making this one of the highest-leverage layers in Codex. About The Role We’re looking for engineers to build the AI systems that make Codex agents dependable in production. The ideal candidate is an agent-systems builder: hands-on across low-level systems and ML workflows, able to debug Codex behavior end to end across the harness, model behavior, inference/runtime stack, GPU fleet, and product surface. You’ll work with research, infrastructure, and product to design agent harness capabilities, run experiments and ablations across the model + system prompt + harness stack, build frameworks for assessing production agent performance, and turn messy failures into durable improvements. What You’ll Do Design and build the core agent harness and execution loop that lets Codex agents interpret model outputs, use tools, execute code, and complete long-horizon tasks safely. Build sandboxing, isolation, orchestration, state, and workflow infrastructure for agents operating in real development environments. Develop evaluation, experimentation, and debugging systems that distinguish harness issues, model behavior, inference/runtime issues, and product failures. Run ablations across prompts, model-facing interfaces, context construction, tool-use strategies, and harness behavior to

About the Team OpenAI’s Hardware organization develops AI-native silicon and system-level solutions for the unique demands of advanced AI workloads. Building on efforts like Jalapeño, the team is developing future generations of AI-native silicon and tightly integrated systems to power the next generation of frontier models. By co-designing chips, systems, tools, and methodologies, the team helps deliver faster, more efficient, and production-ready hardware for OpenAI’s supercomputing platform. About the Role On the Accelerators team, you will help OpenAI evaluate and bring up new compute platforms that can support large-scale AI training and inference. Your work will range from prototyping system software on new accelerators to enabling performance optimizations across our AI workloads. You’ll work across the stack, collaborating with both hardware and software aspects - working on kernels, sharding strategies, scaling across distributed systems, and performance modeling. You'll help adapt OpenAI's software stack to non-traditional hardware and drive efficiency improvements in core AI workloads. This is not a compiler-focused role, rather bridging ML algorithms with system performance - especially at scale. In this role, you will: Prototype and enable OpenAI's AI software stack on new, exploratory accelerator platforms. Optimize large-scale model performance (LLMs, recommender systems, distributed AI workloads) for diverse hardware environments. Develop kernels, sharding mechanisms, and system scaling strategies tailored to emerging accelerators. Collaborate on optimizations at the model code level (e.g. PyTorch) and below to enhance performance on non-traditional hardware. Perform system-level performance modeling, debug bottlenecks, and drive end-to-end optimization. Work with hardware teams and vendors to evaluate alternatives to existing platforms and adapt the software stack to their architectures. Contribute to runtime improvements, compute/communication over

About the Team OpenAI’s Hardware organization develops AI-native silicon and system-level solutions for the unique demands of advanced AI workloads. Building on efforts like Jalapeño, the team is developing future generations of AI-native silicon and tightly integrated systems to power the next generation of frontier models. By co-designing chips, systems, tools, and methodologies, the team helps deliver faster, more efficient, and production-ready hardware for OpenAI’s supercomputing platform. About the Role You will build the model runtime within the inference engine that executes complex, frontier models at scale on OpenAI’s custom silicon. The runtime will sit between models running on the hardware and the upper layers of the cluster serving software stack, translating demanding inference workloads into efficient execution while optimizing for throughput, latency, utilization, and reliability. You will work across model architecture, distributed systems, compilers, kernels, and silicon to design a production-grade runtime comparable in ambition to systems such as vLLM and SGLang, but customized and optimized for OpenAI’s AI accelerator. Your work will shape how new model capabilities map onto the platform and how quickly custom silicon can deliver meaningful performance in production. In this role, you will: Design and implement the LLM inference runtime for frontier models running on custom silicon. Build scheduling, continuous batching, memory management, KV-cache management, and execution orchestration for high-performance inference. Develop distributed execution strategies across chips, hosts, and racks, including model partitioning, communication, and synchronization. Optimize end-to-end latency, throughput, memory efficiency, and hardware utilization across diverse model architectures and serving workloads. Partner with kernel, compiler, architecture, and silicon teams to co-design interfaces and remove performance bottlenecks across the stack. Enable new

About the Team The GPT Infrastructure team builds systems that turn advances in model inference and optimization into reliable production capabilities. We enable OpenAI workloads to be qualified and optimized across new accelerator platforms without requiring a one-off port and tuning effort for every hardware target. Our work spans distributed systems, model execution, compilers and runtimes, performance engineering, secure partner integrations, evaluation systems, and developer tooling. We build the infrastructure that makes optimization workflows automated, reproducible, and trustworthy. About the Role We are seeking a software engineer to help build the platform that qualifies and optimizes inference workloads across heterogeneous compute environments. You will develop both OpenAI-hosted services and secure partner-side software for running long-lived optimization workflows. These workflows generate candidate kernels, runtime configurations, and serving-stack changes; compile and execute them on target hardware; verify their correctness; measure their performance; and use the results to guide further optimization. You will work across model architecture, distributed execution, compilers, runtimes, networking, and accelerator systems. A central part of the role is turning research prototypes and one-off hardware bring-up efforts into reliable, reusable infrastructure with clear contracts, reproducible results, strong observability, and well-defined security boundaries. Key Responsibilities Design, build, and operate APIs and control-plane services for long-running workload qualification and optimization campaigns, including scheduling, retries, checkpointing, resource budgets, and observability. Build secure partner-side execution and evaluation software that can compile, run, verify, profile, and benchmark candidate artifacts on accelerator hardware. Integrate model workloads, hardware profiles, compiler toolchains, runtimes, serving engines, and distributed-exe

About the Team OpenAI’s Hardware organization develops silicon and system-level solutions designed for the unique demands of advanced AI workloads. The team is responsible for building the next generation of AI-native silicon while working closely with software and research partners to co-design hardware tightly integrated with AI models. In addition to delivering production-grade silicon for OpenAI’s supercomputing infrastructure, the team also creates custom design tools and methodologies that accelerate innovation and enable hardware optimized specifically for AI. About the Role As an Engineer on our hardware optimization and co-design team, you will co-design future hardware from different vendors for programmability and performance. You will work with our kernel, compiler and machine learning engineers to understand their unique needs related to ML techniques, algorithms, numerical approximations, programming expressivity, and compiler optimizations. You will evangelize these constraints with various vendors to develop and influence future hardware architectures towards efficient training and inference on our models. If you are excited about efficiently distributing a large language model across devices, dealing with and optimizing system-wide/rack-wide networking bottlenecks and eventually tailoring the compute pipe and memory hierarchy of the hardware platform, simulating workloads at different abstractions and working closely with our partners, this is the perfect opportunity! This role is based in San Francisco, CA. We use a hybrid work model of 3 days in the office per week and offer relocation assistance to new employees. Key Responsibilities Co-design future hardware for programmability and performance with our hardware vendors Assist hardware vendors in developing optimal kernels and add support for it in our compiler Develop performance estimates for critical kernels for different hardware configurations and drive decisions on compute core and memory h

About the Team Our mission at OpenAI is to discover and enact the path to safe, beneficial AGI. To do this, we believe that many technical breakthroughs are needed in generative modeling, reinforcement learning, large-scale optimization, active learning, and other areas. The team builds the performance-critical systems that allow OpenAI's models to run efficiently across a diverse set of AI accelerators. We work across the inference stack, from low-level kernels and compilers through model execution, to unlock the full capabilities of the underlying hardware. About the Role As a Software Engineer, Trainium, you will help bring OpenAI's inference workloads to AWS Trainium and build the software stack required to run cutting-edge frontier models efficiently on the platform. This is a deeply technical, cross-stack role spanning kernels, compilers, and model execution. You will work on the systems needed to support OpenAI's inference stack on Trainium, including developing and optimizing high-performance kernels, improving compiler support, and enabling efficient execution of the model forward pass. You'll work closely with engineers across inference, compilers, kernels, and ML systems to identify performance bottlenecks and build the software needed to take full advantage of Trainium. The work may range from low-level hardware-specific optimization to compiler and runtime improvements to integrating new model architectures into the inference stack. If you enjoy working at the intersection of ML systems, compilers, kernels, and accelerator hardware, this role is for you. We're looking for engineers who are self-directed, comfortable operating across abstraction layers, and excited to solve challenging performance problems for frontier-scale AI systems. In This Role, You Will Build and optimize OpenAI's inference stack for AWS Trainium. Develop high-performance kernels for critical model operations and workloads. Extend and improve compiler support to efficiently target