Full-Time
Network Engineer
Posted on 7/29/2024
Cerebras
201-500 employees
Develops AI acceleration hardware and software
Mid
Toronto, ON, Canada + 1 more
More locations: Sunnyvale, CA, USA
- Engineering degree, or a related technical discipline or equivalent experience
- Expert Knowledge of IB/RDMA/RoCE Networks
- 4+ years of experience working on networks supporting large scale training workloads
- Understanding of RDMA congestion control mechanisms on IB and RoCE Networks.
- Hands on experience working with state-of-the-art network equipment and vendors (e.g. Broadcom, Mellanox)
- Experience coding in languages like Python, C++, Go, etc
- Experience in network automation software leveraging software defined networking principles.
- Developed or modified network telemetry and automation tools to make efficient use of infrastructure and resources, related to performance, operation, testing, and incident management.
- Experience in designing, deploying and operating networks at scale. Built and managed large scale data center networks or experience with building transports for large scale networks.
- Design, develop, test and operate networking systems to support large scale AI training/inference jobs.
- Develop and deploy numerous technologies and network topologies in order to evolve and scale our AI networks.
- Work closely with our hardware, software and sourcing teams to develop new networking solutions and influence the future of networking and its associated infrastructure
- Define and develop optimized network monitoring systems
- Software modelling of network architecture, ML applications and other building blocks to realistically simulate end-to-end performance at scale
- Be oncall to learn from real world production challenges and take the lessons to improve current and future generation products.
Cerebras Systems specializes in accelerating artificial intelligence (AI) processes with its CS-2 system, which is designed to replace traditional clusters of graphics processing units (GPUs) used in AI computations. The CS-2 system simplifies the complexities of parallel programming, distributed training, and cluster management, making AI tasks more efficient. Clients from various sectors, including pharmaceuticals, government research labs, healthcare, finance, and energy, utilize the CS-2 to achieve faster results in critical applications like cancer drug response prediction. Cerebras generates revenue by selling its proprietary hardware and software solutions, including the CS-2 systems and associated cloud services. The company's goal is to streamline AI research and development, enabling clients to reduce costs and improve the speed of AI training and inference.
Company Stage
Series F
Total Funding
$700.4M
Headquarters
Sunnyvale, California
Founded
2016
6 month growth
↑ 6%1 year growth
↑ 16%2 year growth
↓ -1%
Simplify's Take
What believers are saying
- Collaboration with Dell expands Cerebras' reach in enterprise generative AI projects.
- The WSE-3 processor attracts clients seeking cutting-edge AI technology.
- Cerebras' IPO filing suggests potential for increased capital for R&D and expansion.
What critics are saying
- Reliance on high-profile clients like GlaxoSmithKline poses a risk if they switch.
- Competition with Nvidia may lead to aggressive pricing and reduced profit margins.
- Dependency on Dell's distribution channels could influence Cerebras' product positioning.
What makes Cerebras unique
- Cerebras' WSE-3 chip has 1.4 trillion transistors, surpassing competitors in AI hardware.
- The CS-2 system replaces traditional GPU clusters, simplifying AI computations significantly.
- Cerebras' AI inference service offers unmatched performance and cost efficiency in the market.
Help us improve and share your feedback! Did you find this helpful?