Full-Time
Software Engineer
Posted on 6/27/2024
Clockwork Systems
1-10 employees
Advanced clock synchronization for distributed systems
No salary listed
Palo Alto, CA, USA
In Person
 Software Engineering (1) |
|---|
- A degree in computer science, electrical engineering or related field
- Fluency in at least one of the following languages: C, C++, Golang, Java, Python
- Excellent communication skills
- Excellent debugging and problem solving skills
- Use your coding skills and distributed system knowledge to contribute directly to the design and build of high performance, reliable and scalable systems.
- Operating systems expertise (e.g. threads, scheduling, process synchronization, kernel programming)
- Computer networks expertise (e.g. networking stacks and protocols, queuing theory, congestion control)
- Experience designing, building and deploying large-scale distributed systems
- Full-stack development experience
Clockwork Systems provides clock synchronization technology for mission-critical distributed systems, ensuring precise timing across operations. Its solutions run in both cloud and on-premises environments and are delivered through software licensing, subscriptions, and professional services. The company differentiates itself with deep timing expertise and end-to-end timing across networks, framed by tools like Latency Sensei for cloud latency monitoring. Its goal is to help customers achieve reliable, accurate synchronization to boost performance and reduce timing-related issues in time-sensitive applications.
Company Size
1-10
Company Stage
Early VC
Total Funding
$41.6M
Headquarters
Palo Alto, California
Founded
2018
Simplify's Take
What believers are saying
- Raised $20.6M Series A in 2022 led by NEA for FleetIQ launch.
- Serves hyperscalers, banks, pharma labs optimizing model training.
- Hired NetApp exec Suresh Vasudevan as CEO in 2025.
What critics are saying
- NVIDIA software stack bundles GPU management, blocks penetration in 12-24 months.
- Hyperscalers like Meta, Google build proprietary fabrics in 18-36 months.
- Customer concentration risks 30-50% revenue loss from one cloud exit.
What makes Clockwork Systems unique
- FleetIQ delivers microsecond visibility into GPU clusters for AI workloads.
- Software-driven fabric runs on NVIDIA, AMD, InfiniBand, RoCE, Ethernet.
- Stanford spinout founded 2018 extends clock sync to AI training.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Competitive Salary
Company News
Clockwork.io has launched TorchPass Workload Fault Tolerance, a software solution that eliminates costly GPU training failures through Live GPU Migration technology. The system allows AI training workloads to continue running through hardware failures, network disruptions and node crashes without requiring checkpoint restarts. The company claims TorchPass can save over $6 million annually in a typical 2,048-GPU deployment by reducing wasted training progress by 95%. In large clusters, it cuts lost time from approximately three hours per day to under ten minutes. Independent testing by SemiAnalysis found TorchPass delivered faster fault-tolerant performance than standard checkpoint-restart approaches and higher Model FLOPs Utilisation than leading open-source alternatives. The solution typically completes recovery in approximately three minutes whilst training continues uninterrupted. TorchPass is now available as part of Clockwork.io's FleetIQ platform.
Clockwork Raises $20.57 Million in Funding
Stanford spinout Clockwork has raised $20.6 million, led by NEA with participation from notable investors, to address AI's GPU inefficiency. The funding coincides with the launch of FleetIQ, a software solution aimed at enhancing GPU performance by improving communication between GPUs, clusters, and clouds. This innovation seeks to reduce crashes, shorten restarts, and increase utilization rates, making AI infrastructure more efficient and sustainable.