xAiSenior Engineer - Post-training Infrastructure
Seeking experienced software engineers to design, implement, and optimize large-scale distributed training systems for LLMs, particularly in RL and Agent.
About xAI
xAI’s mission is to create AI systems that can accurately understand the universe and aid humanity in its pursuit of knowledge.
Our team is small, highly motivated, and focused on engineering excellence. This organization is for individuals who appreciate challenging themselves and thrive on curiosity.
We operate with a flat organizational structure. All employees are expected to be hands-on and to contribute directly to the company’s mission. Leadership is given to those who show initiative and consistently deliver excellence. Work ethic and strong prioritization skills are important.
All engineers and researchers are expected to have strong communication skills. They should be able to concisely and accurately share knowledge with their teammates.
About the Role
xAI is seeking experienced software engineers to design, implement and optimize large-scale distributed training systems for LLM, particularly in the areas of Reinforcement Learning (RL) and Agent. The training system needs to be robust, fast, and reasonably flexible to support state-of-the-art research.
Focus
- Design and implement distributed RL training systems
- Profile, debug and optimize system performance
- Software and algorithm co-design with researchers
Ideal Experiences
Built scalable training framework for AI models in HPC clusters, including but not limited to
- Scalable orchestration framework and tools
- Reinforcement learning framework consists of asynchronous training, inference, simulation and more components
Experience in configuring and troubleshooting operating systems for maximum performance.
Experience in building high-performance sandboxes, virtual machines and simulations.
Typical problems you will deal with
- We have a new algorithm that requires a certain computation / communication pattern that the current system does not support or is very inefficient to handle. How should we refactor or improve the current system to support this algorithm?
- We have a new pre-training model, which has very different traits compared to the previous generation. As a result, the current system does not quite work. How should we refactor or improve the current system?
- The amount of FLOPs / HBM bandwidth / throughput the current system can achieve is only “XXX”, which is only “YY%” of the theoretical limit. How much can we improve?
- The training run restarts / crashes every “XXX” hours. What are the root causes of these restarts / crashes? Is there any way to reduce the errors?
Tech Stack
- Python / Rust / C++
- JAX and PyTorch
- CUDA and NCCL
Interview Process
After submitting your application, the team reviews your CV and statement of exceptional work.
If your application passes this stage, you will be invited to a 15-minute phone interview during which a member of our team will ask some basic questions.
If you clear the initial phone interview, you will enter the main process, which consists of four technical interviews:
- Coding assessment in a language of your choice.
- Two systems hands-on: Demonstrate practical skills in live problem-solving sessions that involve both system design and coding.
- Meet the Team: Present your past exceptional work and your vision with xAI to a small audience.
Our goal is to finish the main process within one week. We don’t rely on recruiters for assessments. Every application is reviewed by a member of our technical team. All interviews will be conducted via Google Meet.
Location
The role is based in the Bay Area [San Francisco and Palo Alto]. Candidates are expected to be located near the Bay Area or open to relocation.
xAI is an equal opportunity employer and does not unlawfully discriminate based on race, color, religion, ethnicity, ancestry, national origin, sex (including pregnancy, childbirth, or related medical conditions), sexual orientation, gender, gender identity, gender expression, age, disability, medical conditions, genetic information, marital status, military or veteran status, or any other applicable legally protected characteristics.
Qualified applicants with arrest or conviction records will be considered for employment in accordance with all applicable federal, state, and local laws, including the San Francisco Fair Chance Ordinance, Los Angeles County Fair Chance Ordinance for Employers, and the California Fair Chance Act.
For Los Angeles County (unincorporated) Candidates:
xAI reasonably believes that criminal history may have a direct, adverse and negative relationship on the following job duties, potentially resulting in the withdrawal of a conditional offer of employment:
- Access to information technology systems and confidential information, including proprietary and trade secret information, and/or user data;
- Interacting with internal and/or external clients and colleagues; and
- Exercising sound judgment.